Heart UK

Dwarkesh Podcast

Deeply researched interviews <br/><br/><a href="https://www.dwarkesh.com?utm_medium=podcast">www.dwarkesh.com</a>

Dwarkesh Patel

Ajeya Cotra is a researcher at METR, where she works on threat modeling for loss-of-control risks from advanced AI. Before that, she led the technical AI safety program at what is now Coefficient Giving.She is one the three authors of METR and Redwood Research’s “Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident”.We go through not only what she and her coauthors discovered during this investigation, but what it means for how we should train future, smarter AIs which might be involved in the process of recursive self-improvement.Watch on YouTube; read the transcript.Sponsors* Jane Street’s ML engineering internships start with an intense four-day bootcamp: PyTorch, autograd, writing kernels, profiling workloads… all the things that Jane Street engineers need to know for their daily work. After that, interns tackle real projects, things the firm actually wants in its codebase. If you want to apply, or if you want to watch my recent conversation with Axel, one of Jane Street’s ML engineers, go to janestreet.com/dwarkesh* Cursor, which is now part of SpaceX, noticed that their MoE layers were eating more than half of total training time. So they wrote and open-sourced Mixture-of-Kittens, which is a custom megakernel for training MoE models on NVL72s. This kernel sped up an end-to-end run across 512 GPUs by 1.4x, from about 760 to over 1000 tokens per second per GPU. If you want to read more about the ML research that Cursor and SpaceX are doing, go to cursor.com/dwarkesh* Antithesis hands you (or your agents) a bug’s root cause so you can avoid days of manual debugging. If your test run crashes, Antithesis rewinds, branches off hundreds of slightly varied rollouts, and checks in how many of them the crash still appears. Then it rewinds further and does this all again. As Antithesis rewinds, it eventually finds the spot where the frequency of the crash plummets: that’s where the root cause lives! If you want to see it in action, go to antithesis.com/dwarkeshTimestamps(00:00:00) - Agents get kicked off(00:06:45) - Self-sacrificing behavior(00:13:43) - Potemkin villages(00:23:27) - The Hugging Face attack(00:35:23) - The slopvestigation(00:52:02) - Understanding the AI's motives(01:05:31) - The actual dangers of anthropomorphizing(01:14:30) - What smarter models might do(01:30:29) - The implications for recursive self-improvement(01:38:10) - Is this the case for open source?(01:53:04) - How do we prevent this in the future?(02:15:58) - The clearest warning shot we might ever get This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Ajeya Cotra – Inside the OpenAI agent swarm that hacked Hugging Face

This is a video recording of a post I wrote last week. You can read the original here. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

The rise and fall of agent civilizations

Had a lot of fun chatting again with my twin brother Dylan Patel.We went through lab economics over the next few years - the shift from inference to training as RSI draws near; and how Anthropic and OpenAI are on track to control most of the world’s usable FLOPs within the next few years (because they can monetize compute better and thus outbid everyone).And then we discuss whether the >$10T of total AI capex we’ll see by the end of the decade will cause a sovereign debt crisis, where hyperscaler debt raises interest rates, drives non-AI exposed countries into bankruptcy, and crashes non-AI equities.One question we weren’t able to resolve is whether there’s anything that can counter all the forces barrelling towards centralization in this industry - the economies of scale in training, the scarcity of compute, and eventually continual learning and RSI.Watch on YouTube; read the transcript.Sponsors* Grok Bot has been quite helpful with my search for a new editor. I created a recruiter bot and described the type of editor I was looking for. That bot then spun up a handful of subagents that combed through my emails and X DMs, read the end credits of various documentaries I like, and figured out who edits for some of my favorite YouTubers. It took all of those results, and then delivered me a shortlist of candidates that matched my criteria. Try Grok Bot for yourself at x.ai/bot* Antithesis lets you add time travel to your software testing toolkit. Since the Antithesis platform is fully deterministic, everything that happens inside of it is perfectly reproducible. So if your software crashes, you can rewind to the exact right moment, freeze time, and investigate. Or you can test different hypotheses by perturbing the system: kill a node or disable a feature, see what happens, then reset the trajectory and try something else. Learn more at antithesis.com/dwarkesh* Jane Street is hiring for two separate ML internships right now, one focused primarily on research and one focused on engineering. In both cases, interns are expected to contribute to real work, not contrived exercises: one common project is adapting a frontier LLM paper to financial markets, which tend to come with a ton of different gnarly challenges. Importantly, you don’t need any finance background to apply. 2027 applications are open now at janestreet.com/dwarkeshTimestamps(00:00:00) – Two labs will soon control most of the world’s compute(00:07:01) – $6 billion in fab capex enables $1t+ of end revenue(00:13:08) – Compute prices will rise if the labs outbid everyone(00:18:22) – Which layer will capture most of the surplus?(00:25:40) – What could slow down progress?(00:29:43) – Labs are shifting compute from inference to R&D(00:33:27) – China gets less than 10% of new compute, but its labs need less(00:48:48) – Will AI cause a sovereign debt crisis?(01:07:52) – Will the world’s future workforce belong to a few companies? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Dylan Patel – Anthropic & OpenAI will have most of the world’s compute by 2028

Ryan Greenblatt is the Chief Scientist at Redwood Research, where he works on technical AI safety research. He's also lead author on the "Alignment faking in Large Language Models", and is currently working on a third party investigation into the OpenAI/HuggingFace incident. In my opinion, he's one of the most interesting thinkers on the future of AI.Had him on to discuss/debate recursive self-improvement. This might be the most important question in the world right now – whether within a year or so of achieving human-level intelligence, you slingshot towards having 10s of billions of superintelligences, each of which is dramatically more competent than human experts across all fields.I’ve historically been skeptical of this possibility. My intuition has been that we will end up significantly bottlenecked by not only compute scaling but human expert data, which I think underlies most of the AI progress today.If, because of RSI, we got a jump as big as GPT-3 to a Mythos (i.e. 6 years of AI progress) within a single year of achieving AGI, then the thing we get there at the end of that year is definitively and wildly superhuman.We hashed it out, and I think Ryan made a pretty good case that this kind of speedup is plausible. FWIW, Ryan’s median for when we automate AI R&D is 2031.We then discussed the alignment implications of this scenario. Who should these superintelligences be aligned to? In the future, our capacity to steward our votes and our capital, and to make sense of what’s happening in the world, will all be titrated by superintelligences. And I worry that specs like the Claude Constitution are not shaping these ASIs to truly be my personal advocates and guardian angels.And can we get them aligned to anything in the first place? Ryan and I had a long debate about whether the kind of reward hacking we saw with the OAI/Hugging Face hack extrapolates to superintelligences that would team up to literally take over the world.The first piece of advice you get when you’re learning to drive is that it will go much smoother if you look at the horizon instead of directly in front of your tires. And so it is with the trajectory of AI. Hope you enjoy!Watch on YouTube; read the transcript.Sponsors* Antithesis is a software testing platform that finds the failures no human or AI could ever anticipate. It runs thousands of copies of your code inside a fully deterministic computer, injecting faults and steering each trajectory toward the most insidious bugs. This lets you find critical issues in minutes rather than waiting months for your users to uncover them. Learn more at antithesis.com/dwarkesh* Jane Street’s back with a new puzzle. They designed an ASIC and sent me the final masks… but they didn’t tell me what the chip actually does. So that’s the challenge: reverse engineer the circuit and figure out the chip’s purpose. Jane Street has a bunch of swag ready to send to the most creative solutions, and they’re also planning to feature the top write-ups in a blog post. Download the files and get started at janestreet.com/dwarkesh* Cursor and SpaceX recently released Grok 4.5, and I’ve been surprised by just how good the model is. For example, when I tested it against Fable and Sol on a bunch of AI governance questions, all three models gave substantially the same answers, but Grok was faster, more concise, and cheaper. Grok 4.6 is coming soon, but in the meantime, you can try 4.5 at cursor.com/dwarkeshTimestamps(00:00:00) – Is AI R&D verifiable enough to unlock recursive self-improvement?(00:16:52) – Is AI progress bottlenecked by human expert data?(00:34:02) – Flat token prices suggest scaling has been slow(00:39:47) – Skills AI can’t train on: does it even need them?(00:48:07) – Aligned to whom?(01:09:18) – Recent incidents of AIs colluding and deceiving humans(01:19:38) – What could possibly go wrong? A concrete scenario(01:48:02) – From reward hacking to takeover This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Ryan Greenblatt – What happens once AI can automate AI research?

Read the essay here. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

8 Predictions for the Era of Continual Learning

This is a video recording of a post I wrote last week. If you want to read the original you can check it out here.Thanks to Mercury for sponsoring this video. Mercury’s built-in AI, Command, helps me close my books and saves me a bunch of time. At the end of each month, Command categorizes my transactions and provides its rationale for every choice: I just review, fix anything that’s off, and approve... and then Mercury syncs everything to QuickBooks. Get started at mercury.com/command This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Why smarter AI models could drive up compute prices 10x

Adam Brown is back!General relativity is said to be the most beautiful idea the human mind has ever produced. Most of us will never get to fully appreciate its elegance by taking the 20-lecture graduate course Adam taught on it at Stanford. But in this episode, Adam distills the key idea at its heart so clearly and compellingly that even I could keep up lol.At the core of general relativity, Einstein is trying to figure out the principle behind a particular coincidence: that the mass that resists acceleration and the mass that gravity pulls on just happen to be exactly the same. Adam then leads us through the path of insight which Einstein called his “happiest thought.”Then Adam lectures on black holes. First, by showing how even under special relativity you could create a perpetual motion machine if black holes weren’t truly black. And then, by explaining why the observations of an infalling observer and a distant bystander to the black hole would be so radically differentAdam leads Blueshift, the team at Google DeepMind cracking science and reasoning, which gave us the opportunity to discuss at the very end how close we are to AIs that could rediscover general relativity from scratch. Stay till the close for some philosophy of science.Watch on YouTube; read the transcript.Sponsors* Jane Street has traders from all sorts of different backgrounds. For example, I recently got to speak with Jed Thompson, a trader who started his career in particle physics. Jed told me how the habits he built as a physicist (like never running a calculation without first having a good guess at the answer) helped him build good trading intuition. So no matter what field you’re working in right now, your experience may be more applicable than you think. Check out open positions at janestreet.com/dwarkesh* Crusoe gave me early access to their new serverless fine-tuning product, so I decided to try fine-tuning a Dwarkesh-style question generator. Crusoe made this really easy: I just turned my interview transcripts into training data and then kicked off a run – I never had to touch infra or tweak hyperparameters. After training was done, I ran a blind eval with my team: they preferred the fine-tuned model’s proposed questions over my own suggestions about 30% of the time. Serverless fine-tuning goes live next week. Learn more at crusoe.ai/dwarkesh* Cursor’s iOS app lets me kick off real work no matter where I am. For example, recently I was at dinner with friends when I had an idea about how to investigate the past few years of progress in sample efficiency. I pulled out the Cursor app, dumped my thoughts into a voice note, and 15 minutes later, Cursor had cloned the relevant repo, done the necessary analysis, and written up its findings. And now I’m expanding that work into a full write-up. Without the Cursor app, the idea would’ve floated away. Check out the app now at cursor.com/dwarkeshTimestamps(00:00:00) – The coincidence that led Einstein to general relativity(00:16:42) – Gravity is a consequence of curved spacetime, not a force(00:31:46) – Why black holes prevent unlimited energy extraction(00:47:12) – Black holes are the ultimate power plants(01:13:50) – What falling into a black hole would actually feel like(01:18:51) – The three ways we know black holes are real(01:24:21) – The first time we saw gravity bend light(01:29:33) – How far can AI get without experimental evidence? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Adam Brown – A deep but accessible introduction to general relativity

Always so much fun to chat with Grant.AI has been making much faster progress in math than in other fields. As a result, mathematics is showing us, very concretely, what AI progress in other fields will look like. Even within mathematics, there’s a jagged landscape. What does it look like?What is the nature of the most important conceptual breakthroughs in the history of mathematics, and how different are they from what AIs are currently able to do?Does AI (on net) increase or decrease human understanding of the field?How big is the overhang from having AIs systematically try to connect ideas already in the literature?And what advice does Grant have for aspiring mathematicians, coders, and other students who are passionate about fields that are being most transformed upon by AI?Watch on YouTube; read the transcript.Sponsors* Gemini 3.5 Live Translate is what I wished I’d had on my last trip to China. It detects more than 70 languages and translates them in near real-time… and it preserves your original pacing and intonation. If you’re building an app that needs live translation, you should check out Gemini 3.5 Live Translate. Get started at ai.studio/live* Cursor’s harness lets me use models for a huge range of tasks at the podcast. For example, Cursor cuts out the ads from each episode I produce so I can post them on Bilibili. It also helps me prep for interviews — I have a repo full of books and papers that Cursor sorts through to find the exact right file for any given question. Try Cursor yourself at cursor.com/dwarkesh* Jane Street sponsors 3Blue1Brown, so Grant has gotten to spend a lot of time with various Jane Streeters. He actually just recorded an interview with a few of them, so when we sat down for this episode, he told me about some of the things he learned, like how Jane Street keeps their role definitions fuzzy to make sure their people keep learning and growing. Go check out Grant’s full interview at 3b1b.co/janestreetTimestamps(00:00:00) – AI is discovering new proofs. Is that AGI?(00:11:32) – The verification loop on conceptual breakthroughs can be a century long(00:26:12) – Will we understand an AI proof of the Riemann hypothesis?(00:38:08) – Can AI find the hidden bridges between fields?(00:53:48) – Why real-world tasks don’t fit into RL environments(01:07:07) – Good writing requires theory of mind that AI still lacks(01:16:02) – Why learning will still depend on human curation This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Grant Sanderson – AI and the future of math

Read it here.Thanks to Mercury for sponsoring this essay.Mercury has automated basically my entire bill pay process for my business. I just give contractors a dedicated email address, and when they send an invoice, Mercury automatically creates a draft payment for me to review. I no longer have to hunt through my inbox for invoices or deal with messy spreadsheets to track my bills. Mercury handles it all. Learn more at mercury.comTimestamps:(00:00:00) – The big research bet the labs are making(00:02:12) – Grindability is just as important as verifiability(00:06:10) – Will RLVR alone generalize?(00:08:41) – Getting the learning back to the weights(00:15:22) – Dreaming(00:17:23) – What 2027 looks like This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

The next big breakthrough will be AIs learning on the job

Read the transcript here.Thanks to Mercury for sponsoring this essay!Mercury just released a new feature called Command, which gives me AI right in my banking platform. And since I use Mercury to run basically my entire business, Command has access to all the info it needs to get real work done. I can ask it to send invoices, or categorize expenses, or even transfer money… and Command just handles it. Learn more at mercury.com/commandTimestamps:(00:00:00) – What is really driving AI progress?(00:03:11) – Comparing human vs AI sample efficiency(00:08:46) – Does sample efficiency matter? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

The data black hole at the center of AI

Had Ada Palmer back on – this time to talk about Machiavelli, perhaps the most misunderstood thinker of all time.Machiavelli cut his teeth as a high-level diplomat for Florence, a position from which he got to closely observe the most important rulers in Europe at the time, including the ones who were on the path to destroying his dearly beloved Florence.In 1513 the Medici retook control of Florence and, wrongly suspecting Machiavelli of participating in a coup attempt, fired, tortured, and exiled him.Machiavelli could have left exile and worked for any number of different principalities that would have been eager to make use of his talents.Instead, he decided to rot in the countryside and compile his career’s lessons about power, politics, and human nature into a book he dedicated to the very man whose new regime had tortured and exiled him, Lorenzo di Piero de’ Medici.But at least the Medici were in a position to use his insights to defend Florence. Machiavelli the patriot did not want any other hands to touch these books, because those hands, armed further with these lessons, might pose an existential danger to Florence.The closest modern analogy, at least as Machiavelli would have seen it, would be Szilard’s letter warning FDR about the possibility of a nuclear fission bomb.What were those insights? And how were they inspired by Machiavelli’s dangerous diplomatic missions all across Europe, and his extensive reading of antiquity? Watch this episode with Ada Palmer to find out!By the way, Ada is launching a new podcast which I’m very excited about. The first season will be about Machiavelli - a perfect way to dive deeper into the topics we discussed in this episode. Subscribe at Beforecast’s website to be notified of the first episode, subscribe on YouTube, follow her on Patreon, and if you want even more Ada, check out her FixTheNews Podcast episode, and check out her books and more.Watch on YouTube; read the transcript.Sponsors* Cursor recently saved one of my podcast recordings. When a video file from a shoot came out corrupted, I pointed Cursor at it: it recovered the footage on its own, tracking down the right reference file from the file’s metadata and realigning the out-of-sync audio. My whole team now uses Cursor for everyday tasks, not just coding. Get started at cursor.com/dwarkesh* Jane Street’s hiring process has been going viral on Twitter lately. The memes are pretty funny, but I wanted to see what their interviews were actually like. So I had Ricson, one of Jane Street’s ML researchers, walk me through a retired puzzle: he gave me an image dataset where 50% of the files had been corrupted – I had to figure out how to recover them. If you’re interested in these sorts of puzzles, you can find Jane Street’s open roles at janestreet.com/dwarkesh* Crusoe is turning the AI datacenter buildout into an industrial process. At their massive Colorado factory, they assemble Spark units, modular datacenters with power, cooling, and fire suppression built in. They also manufacture specific components in-house to skip the longest lead times. Crusoe has experience running these Spark units on a range of energy sources, including solar and used EV batteries, ensuring they don’t get bottlenecked by grid availability. Learn more at crusoe.ai/dwarkeshTimestamps(00:00:00) – How Florence bargained with Cesare Borgia for survival(00:15:08) – Machiavelli’s analytical innovations(00:23:58) – Why popes became warlords(00:36:13) – Why the common people demanded nepotism(00:47:57) – Cesare Borgia brought terror to rulers and justice to the people(00:57:55) – Art as a proxy for war(01:06:41) – Florence, a city famous in hell(01:15:57) – The Prince was a job application to Machiavelli’s torturers(01:41:39) – During the Renaissance, original ideas had to be couched in antiquity(01:50:44) – Why copyright began with the Inquisition(02:02:12) – Machiavelli wasn’t Machiavellian This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Ada Palmer – Machiavelli is the most misunderstood thinker of all time

Economics of AGI episode w Alex Imas and Phil Trammell.There’s a bunch of important questions about how we deal with AI that only economics can answer.What is the optimal way to tax and redistribute the wealth that will be generated? How should countries not in the AI supply chain index into the gains? Is there any world where inequality doesn’t explode?It might seem like these questions have obvious answers, but the first thing economics teaches you is that your intuitions can often be entirely wrong.It was very helpful to chat through these things with Alex and Phil.Watch on YouTube; read the transcript.SponsorsJane Street invests heavily in turning smart people into exceptional researchers and engineers. In addition to their apprenticeship model, Jane Street runs lectures and bootcamps in their in-office classrooms -- managers clear their teams’ schedules to encourage attendance. If you’d like to work at a place that takes learning this seriously, Jane Street is hiring. Check out their open roles at janestreet.com/dwarkeshGoogle’s Gemini Omni has incredible video editing capabilities -- you can upload a video and have Omni change the background, adjust lighting, or add specific elements. But Omni is also a preview of how future frontier models will be trained -- fully multimodal on both input and output. You can try it yourself in the Gemini app at gemini.google or in Flow at flow.googleCursor used targeted RL with textual feedback to help train their Composer 2.5 model. One of their researchers, Sasha Rush, gave me an impromptu blackboard lecture to explain how this form of on-policy self-distillation works -- I posted the full thing on X. If you want to try Composer 2.5, go to cursor.com/dwarkeshTimestamps(00:00:00) – Will capital share increase?(00:19:36) – Messy Middle scenario(00:25:57) – How to tax and redistribute AI wealth(00:30:02) – Why demand collapse is unlikely(00:39:26) – Human employees would be hard to integrate into the machine economy(00:43:08) – What if some humans (or AIs) value wealth accumulation intrinsically?(01:01:28) – What should developing countries do? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Alex Imas and Phil Trammell – What remains scarce after AGI?

New blackboard lecture with Reiner Pope: how do chips actually work - starting with basic logic gates, and working up to why GPUs, TPUs, FPGAs, and the human brain each look the way they do.Reiner is CEO of MatX, a new chip startup (full disclosure - I’m an angel investor). He was previously at Google, where he worked on software efficiency, compilers, and TPU architecture.Watch this one on YouTube so you can see the chalkboard. Read the transcript.Sponsors* Crusoe was one of only five GPU clouds that made the gold tier in SemiAnalysis' most recent ClusterMAX report. Gold-tier providers like Crusoe delivered 5-15% lower TCO than silver-tier clouds, even with identical GPU pricing. This is because optimizations like early fault detection and rapid node replacement don't necessarily show up in the sticker price, but still matter a ton in the real world. Learn more at crusoe.ai/dwarkesh* Cursor is where I do most of my work—from reading research papers to visualizing technical concepts to coding up internal tools for the podcast. Most recently, I used it to build two different review interfaces for my essay contest, one that anonymizes submissions for scoring and another that lets me see applicants' essays next to their resumes and websites. Whatever you're working on, you should try doing it in Cursor. Get started at cursor.com/dwarkesh* Jane Street let me ask Ron Minsky and Dan Pontecorvo, two senior Jane Streeters, a bunch of questions about how they use AI. We discussed everything from the types of models they're training to how they think about the future of trading to why they're more bullish than ever on hiring technical talent. You can watch the full conversation and learn more about their open positions at janestreet.com/dwarkeshTimestamps00:00:00 – Building a multiply-accumulate from logic gates00:16:31 – Muxes and the cost of data movement00:26:10 – How systolic arrays work00:39:11 – Clock cycles and pipeline registers00:51:51 – FPGAs vs ASICs01:03:25 – Cache vs scratchpad01:07:27 – Why CPU cores are much bigger than GPU cores01:12:00 – Brains vs chips01:15:33 – A GPU is just a bunch of tiny TPUs This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Reiner Pope – Chip design from the bottom up

Eric Jang walks through how to build AlphaGo from scratch, but with modern AI tools.Sometimes you understand the future better by stepping backward. AlphaGo is still the cleanest worked example of the primitives of intelligence: search, learning from experience, and self-play. You have to go back to 2017 to get insight into how the more general AIs of the future might learn.Once he explained how AlphaGo works, it gave us the context to have a discussion about how RL works in LLMs and how it could work better – naive policy gradient RL has to figure out which of the 100k+ tokens in your trajectory actually got you the right answer, while AlphaGo’s MCTS suggests a strictly better action every single move, giving you a training target that sidesteps the credit assignment problem. The way humans learn is surely closer to the second.Eric also kickstarted an Autoresearch loop on his project. And it was very interesting to discuss which parts of AI research LLMs can already automate pretty well (implementing and running experiments, optimizing hyperparameters) and which they still struggle with (choosing the right question to investigate next, escaping research dead ends). Informative to all the recent discussion about when we should expect an intelligence explosion, and what it would look like from the inside.Watch on YouTube. Read the transcript.And check out the flashcards I wrote to retain the insights.Sponsors* Cursor‘s agent SDK let me build a pipeline to generate flashcards for this episode. For each card, I had an agent read the transcript, ingest blackboard screenshots, generate an SVG visual, and run everything through a critic. A durable agent is much better at this kind of work than a chain of LLM calls, and Cursor’s SDK made it easy. Check out the cards at flashcards.dwarkesh.com and get started with the SDK at cursor.com/dwarkesh* Jane Street gave me a real deep-dive tour of one of their datacenters. I got to ask a bunch of questions to Ron Minsky, who co-leads Jane Street’s tech group, and Dan Pontecorvo, who runs Jane Street’s physical engineering team. They were willing to literally pull up the floorboards and take out racks to explain how everything works. Check out the full tour at janestreet.com/dwarkeshTimestamps(00:00:00) – Basics of Go(00:08:17) – Monte Carlo Tree Search(00:32:04) – What the neural network does(01:00:33) – Self-play(01:25:38) – Alternative RL approaches(01:45:47) – Why doesn't MCTS work for LLMs(02:01:09) – Off-policy training(02:12:02) – RL is even more information inefficient than you thought(02:22:16) – Automated AI researchers This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Eric Jang – Building AlphaGo from scratch

David Reich is back.He and collaborator Ali Akbari just published a paper that overturns a long-standing consensus about human evolution — that natural selection has been dormant in our species since the agricultural revolution.By scaling ancient DNA sequencing and developing a new statistical method, they found that selection has actually sped up.Selection went especially bonkers during the Bronze Age (around 3,000 years ago).That’s when gene frequencies for everything from immune function to body fat to intelligence were most in flux.Over the last 10,000 years, selection pushed the genetic predictor of cognitive performance up by roughly a full standard deviation — most of it between 4,000 and 2,000 years ago.After we finished recording, David sketched out on a whiteboard his new heretical model about who the Neanderthals really were. Luckily, I took out my iPhone and managed to record it.He thinks the standard story (that Neanderthals are some separate archaic lineage we interbred with a little) just doesn’t fit the evidence. Instead, he proposes that Neanderthals are essentially genetically-swamped modern humans.A small population somewhere around the Caucasus invented Middle Stone Age technology roughly 300,000 years ago and expanded outward. The ones that moved into Europe interbred with local archaic humans, got genetically swamped, and became Neanderthals. The same expansion went into Africa, met much more diverged archaic Africans, and that mixture became us.This means Neanderthals and modern humans share the same cultural ancestry — the only difference is which archaic humans they mixed with afterward.David is a brilliant and rigorous scholar. It was a real delight to learn from him again.Watch on YouTube; read the transcript.Sponsors* Cursor was super useful as I prepped for this episode. Whenever I had a question, I’d have Cursor kick off a few different models simultaneously and then compare their responses. I found that this led to better results than I could get out of any individual LLM. If you’ve only used Cursor for coding, you should try using it for research. Check it out at cursor.com/dwarkesh* Jane Street uses an internal currency called “hive bucks” to allocate compute through a real-time auction – and anyone can change anyone else’s bids or even kill their jobs! Everyone just trusts each other to act in the firm’s best interest, which is what lets the system work in the first place. If this weird and high-trust culture sounds like your kind of thing, Jane Street’s hiring at janestreet.com/dwarkesh* Crusoe’s ML infra team built fastokens, an open-source tokenizer that delivers a ~9x speedup over Hugging Face and up to 40% faster time-to-first token – on real production workloads! Crusoe achieved these results by parallelizing things and using some clever engineering to handle duplicates without cross-thread coordination. Learn more at crusoe.ai/dwarkeshTimestamps(00:00:00) – Ancient DNA suggests strong selection over last 10,000 years(00:15:45) – Natural selection intensified during the Bronze Age(00:35:02) – Why didn’t evolution max out intelligence?(00:57:21) – Evolution is limited by time, not population size(01:09:02) – Why no farming before the Ice Age?(01:17:13) – The Neanderthal puzzle David can’t stop thinking about(01:54:10) – The methodology behind this breakthrough This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

David Reich – Why the Bronze Age was an inflection point in human evolution

Did a very different format with Reiner Pope - a blackboard lecture where he walks through how frontier LLMs are trained and served.It’s shocking how much you can deduce about what the labs are doing from a handful of equations, public API prices, and some chalk.It’s a bit technical, but I encourage you to hang in there – it’s really worth it.There are less than a handful of people who understand the full stack of AI, from chip design to model architecture, as well as Reiner. It was a real delight to learn from him.Recommend watching this one on YouTube so you can see the chalkboard.Reiner is CEO of MatX, a new chip startup (full disclosure - I’m an angel investor). He was previously at Google, where he worked on software efficiency, compilers, and TPU architecture.Download markdown of transcript here to chat with an LLM.Wrote up some flashcards and practice problems to help myself retain what Reiner taught. Hope it's helpful to you too!Sponsors* Jane Street needs constant access to incredibly low-latency compute. I recently asked one of their engineers, Clark, to talk me through how they meet these demands. Our conversation—which touched on everything from FPGAs to liquid cooling—was extremely helpful as I prepped to interview Reiner. You can watch the full discussion and explore Jane Street’s open roles at janestreet.com/dwarkesh* Google’s Gemma 4 is the first open model that’s let me shut off the internet and create a fully disconnected “focus machine”. This is because Gemma is small enough to run on my laptop, but powerful enough to actually be useful. So, to prep for this interview, I downloaded Reiner’s scaling book, disconnected from wifi, and used Gemma to help me break down the material. Check it out at goo.gle/Gemma4* Cursor helped me turn some notes I took on how gradients flow during large-scale pretraining into a great animation. At first, I wasn’t sure the best way to visualize the concept, but Cursor’s Composer 2 Fast model let me iterate on different ideas almost instantaneously. You can check out the animation in my recent blog post. And if you have something to visualize yourself, go to cursor.com/dwarkeshTimestamps(00:00:00) – How batch size affects token cost and speed(00:32:09) – How MoE models are laid out across GPU racks(00:47:12) – How pipeline parallelism spreads model layers across racks(01:03:37) – Why Ilya said, “As we now know, pipelining is not wise.”(01:18:59) – Because of RL, models may be 100x over-trained beyond Chinchilla-optimal(01:33:02) – Deducing long context memory costs from API pricing(02:04:02) – Convergent evolution between neural nets and cryptography This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Reiner Pope – The math behind how LLMs are trained and served

I asked Jensen about TPU competition, Nvidia’s lock on the ever more bottlenecked supply chain needed to make advanced chips, whether we should be selling AI chips to China, why Nvidia doesn’t just become a hyperscaler, how it makes its investments, and much more. Enjoy!Watch on YouTube; read the transcript.Sponsors* Crusoe’s cloud runs on state-of-the-art Blackwell GPUs, with Vera Rubin deployment scheduled for later this year. But hardware is only part of the story—for inference, Crusoe’s MemoryAlloy tech implements a cluster-wide KV cache, delivering up to 10x faster TTFT and 5x better throughput than vLLM. Learn more at crusoe.ai/dwarkesh* Cursor helped me build an AI co-researcher over the course of a weekend. Now I have an AI agent that I can collaborate with in Google Docs via inline comment threads! And while other agentic coding tools feel like a total black-box, Cursor let me stay on top of the full implementation. You can try my co-researcher out at github.com/dwarkeshsp/ai_coworker, or get started on your own Cursor project today at cursor.com/dwarkesh* Jane Street spent ~20,000 GPU hours training backdoors into 3 different language models, then challenged my audience to find the triggers. They received some clever solutions—like comparing the base and fine-tuned versions and extrapolating any differences to reveal the hidden backdoor—but no one was able to solve all 3. So if open problems like this excite you, Jane Street is hiring. Learn more at janestreet.com/dwarkeshTimestamps(00:00:00) – Is Nvidia’s biggest moat its grip on scarce supply chains?(00:16:25) – Will TPUs break Nvidia’s hold on AI compute?(00:41:06) – Why doesn’t Nvidia become a hyperscaler?(00:57:36) – Should we be selling AI chips to China?(01:35:06) – Why doesn’t Nvidia make multiple different chip architectures? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Jensen Huang – TPU competition, why we should sell chips to China, & Nvidia’s supply chain moat

Really enjoyed chatting with Michael Nielsen about how we recognize scientific progress.It's especially relevant for closing the RL verification loop for scientific discovery.But it's also a surprisingly mysterious and elusive question when you look at the history of human science.We approach this question stories like Einstein (who claimed that he hadn't even heard of the famous Michelson-Morley experiment, which is supposed to have motivated special relativity, until after he had come up with the theory), Darwin (why did it take till 1859 to lay out an idea whose essence every farmer since antiquity must have observed?), Prout (how do you recognize that isotopes exist if you cannot chemically separate them?), and many others.The verification loop on scientific ideas is often extremely long and weirdly hostile. Ancient Athenians dismissed Aristarchus's heliocentrism in the 3rd century BC because it would imply that the stars should shift in the sky as the Earth orbits the sun. The first successful measurement of stellar parallax was in 1838. That's a 2,000-year verification loop.But clearly human science is able to make progress faster than raw experimental falsification/verification would imply, and in cases where experiments are very ambiguous. How?Michael has some very deep and provocative hypotheses about the nature of progress. One I found especially thought-provoking is that aliens will likely have a VERY different science + tech stack than us. Which contradicts the common sense picture of a linear tech tree that I was assuming. And has some interesting implications about how future civilizations might trade and cooperate with each other.Watch on Youtube; read the transcript.Sponsors* Labelbox researchers built a new safety benchmark. Why? Well, current safety benchmarks claim that attacks on top models are successful only a few percent of the time, but the prompts in those benchmarks don’t reflect how real bad actors actually write. You can read Labelbox’s research here. If this could be useful for your work, reach out at labelbox.com/dwarkesh* Mercury has an MCP that lets you give an LLM access to your full transaction history, including things like attached receipts and internal notes. I just used it to categorize my 2025 transactions, and it worked shockingly well. Modern functionality like this is exactly why I use Mercury. Learn more at mercury.com* Jane Street’s ML engineers presented some of their GPU optimization workflows at GTC, showing how they use CUDA graphs, streams, and custom kernels to shave real time off their training runs. You can watch the full talk here. And they open-sourced all the relevant code here. If this kind of stuff excites you, Jane Street is hiring — learn more at janestreet.com/dwarkeshTimestamps(00:00:00) – How scientific progress outpaces its verification loops(00:17:51) – Newton was the last of the magicians(00:23:26) – Why wasn’t natural selection obvious much earlier?(00:29:52) – Could gradient descent have discovered general relativity?(00:50:54) – Why aliens will have a different tech stack than us(01:15:26) – Are there infinitely many deep scientific principles left to discover?(01:26:25) – What drew Michael to quantum computing so early?(01:35:29) – Does science need a new way to assign credit?(01:43:57) – Prolificness versus depth(01:49:17) – What it takes to actually internalize what you learn This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Michael Nielsen – How science actually progresses

We begin the episode with the absolutely ingenious and surprising way in which Kepler discovered the laws of planetary motion.People sometimes say that AI will make especially fast progress at scientific discovery because of tight verification loops.But the story of how we discovered the shape of our solar system shows how the verification loop for correct ideas can be decades (or even millennia) long.During this time, what we know today as the better theory can actually make worse predictions.And the reasons it survives this epistemic hell is some mixture of judgment and heuristics that we don’t even understand well enough to actually articulate, much less codify into an RL loop. Hope you enjoy!Watch on YouTube; read the transcript.Sponsors- Jane Street loves challenging my audience with different creative puzzles. One of my listeners, Shawn, solved Jane Street’s ResNet challenge and posted a great walk-through on X. If you want to try one of these puzzles yourself, there’s one live now at janestreet.com/dwarkesh.- Labelbox can get you rubric-based evals, no matter your domain. These rubrics allow you to give your model feedback on all the dimensions you care about, so you can train how it thinks, not just what it thinks. Whatever you’re focused on—math, physics, finance, psychology or something else—Labelbox can help. Learn more at labelbox.com/dwarkesh.- Mercury just released a new feature called Insights. Insights summarizes your money in and out, showing you your biggest transactions and calling out anything worth paying attention to. It’s a super low-friction way to stay on top of your business. Learn more at mercury.com/insights.Timestamps(00:00:00) – Kepler was a high temperature LLM(00:11:44) – How would we know if there’s a new unifying concept within heaps of AI slop?(00:26:10) – The deductive overhang(00:30:31) – Selection bias in reported AI discoveries(00:46:43) – AI makes papers richer and broader, but not deeper(00:53:00) – If AI solves a problem, can humans get understanding out of it?(00:59:20) – We need a semi-formal language for the way that scientists actually talk to each other(01:09:48) – How Terry uses his time(01:17:05) – Human-AI hybrids will dominate math for a lot longer This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Terence Tao – Kepler, Newton, and the true nature of mathematical discovery

Dylan Patel, founder of SemiAnalysis, provides a deep dive into the 3 big bottlenecks to scaling AI compute: logic, memory, and power.And walks through the economics of labs, hyperscalers, foundries, and fab equipment manufacturers.Learned a ton about every single level of the stack. Enjoy!Watch on YouTube; read the transcript.Sponsors* Mercury has already saved me a bunch of time this tax season. Last year, I used Mercury to request W-9s from all the contractors I worked with. Then, when it came time to issue 1099s this year, I literally just clicked a button and Mercury sent them out. Learn more at mercury.com.* Labelbox noticed that even when voice models appear to take interruptions in stride, their performance degrades. To figure out why, they built a new evaluation pipeline called EchoChain. EchoChain diagnoses voice models’ specific failure modes, letting you understand what your model needs to truly handle interruptions. Check it out at labelbox.com/dwarkesh.* Jane Street is basically a research lab with a trading desk attached – and their infrastructure backs this up. They’ve got tens of thousands of GPUs, hundreds of thousands of CPU cores, and exabytes of storage. This is what it takes to find subtle signals hidden deep within noisy market data. If this sounds interesting, you can explore open positions at janestreet.com/dwarkesh.Timestamps(00:00:00) – Why an H100 is worth more today than 3 years ago(00:24:52) – Nvidia secured TSMC allocation early; Google is getting squeezed(00:34:34) – ASML will be the #1 constraint for AI compute scaling by 2030(00:55:47) – Can't we just use TSMC's older fabs?(01:05:37) – When will China outscale the West in semis?(01:16:01) – The enormous incoming memory crunch(01:42:34) – Scaling power in the US will not be a problem(01:54:44) – Space GPUs aren't happening this decade(02:14:07) – Why aren't more hedge funds making the AGI trade?(02:18:30) – Will TSMC kick Apple out from N2?(02:24:16) – Robots and Taiwan risk This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Dylan Patel — Deep dive on the 3 big bottlenecks to scaling AI compute

Read the full essay here: https://www.dwarkesh.com/p/dow-anthropicTimestamps(00:00:00) - Anthropic vs The Pentagon(00:04:16) - The overhangs of tyranny(00:05:54) - AI structurally favors mass surveillance(00:08:25) - Alignment...to whom?(00:13:55) - Coordination not worth the costs This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

The most important question nobody's asking about AI

Renaissance history is so much wilder and weirder than you would have expected. Very fun chatting with Ada Palmer (historian, novelist, and composer based at the University of Chicago).Some especially fascinating things I learned from the conversation and her excellent book, Inventing the Renaissance:Not only did Gutenberg go bankrupt in the 1450s (after inventing the printing press), but so did the bank that foreclosed on him, and so did his apprentices. This is because paper was still very expensive, and so you had to make this big upfront CAPEX decision to print a batch of 300 copies of a book - say the Bible. But he’s in a small landlocked German town where only priests are allowed to read the Bible - so he sells maybe 7 copies. It’s only when this technology ends up in Venice, where you can hand 10 copies to each of 30 ship captains going to 30 different cities, that it starts taking off.Speaking of which, the printing revolution wasn’t just one single discrete event, just as the computer revolution has been this whole century of going from mainframes -> personal computers -> phones -> social media, each with different and accelerating social impact. Books came first, but they’re slow to print, and made in small batches. The real revolution is pamphlets - much faster, much harder to censor. Pamphlet runners are how you can have Luther’s 95 Theses go from Wittenberg to London in 17 days.So much other wild stuff from this episode. For example, did you know that the largest and best-funded experimental laboratory in 17th century Europe was very likely the Roman one run by inquisitors? Ada jokes that the Inquisition accidentally invented peer review. The focus of the Inquisition is really misunderstood - it was obsessed with catching dangerous new heretics like Lutherans and Calvinists - it only executed one person for doing science.And this leads Ada to make an observation that I think is really wise: the authorities and censors are always worried about the exact wrong things given 20/20 hindsight. When Inquisition raids an underground bookshop during the French Enlightenment, they don’t mind the Rousseau, Voltaire, and Encyclopédie, but they lose their minds about some Jansenist treatises about the technical nature of the Trinity.More broadly, a lesson for me from this episode is that it’s just really hard to shape history in the specific way that you want to impact things. One of the most famous medieval scholars is this guy Petrarch. He survives the Black Death in the 1340s, watches his friends die to plague and bandits, and says: our leaders are selfish and terrible, we need to raise them on the Roman classics so they’ll act like Cicero. So Europe pours money into finding ancient manuscripts, building libraries, and educating princes on classical virtues. Those princes grow up and fight bigger, nastier wars than ever before with new deadlier technology. And this, combined with greater urbanization and endemic plague, results in European life expectancy decreasing from 35 in the medieval period to 18 during the Renaissance (the period which we in retrospect think of as a golden age but which many people living through it thought of as the continuation of the dark ages that had persisted since the fall of Rome).Anyways, the libraries Petrarch inspires stick around, the printing press makes them accessible to everyone, and 200 years later a generation of medical students is reading Lucretius and asking “what if there are atoms and that’s how diseases work?” which eventually leads to germ theory, vaccines, and a cure for the Black Death (Ada has longer more involved explanation of how cosplaying the Romans results through a series of many steps to the scientific revolution). Petrarch wanted to produce philosopher-kings that shared his values. Instead he created a world that doesn’t share his values at all but can cure the disease that destroyed his.Watch on YouTube; read the transcript.Sponsors* Jane Street is still waiting on someone to solve their backdoor puzzle… They’re accepting submissions until April 1st and have set aside $50,000 for the best attempts. Separately, applications are live for Jane Street’s summer ML internships in NY, London, and Hong Kong. Go check all of this out at janestreet.com/dwarkesh.* Labelbox can help ensure your agents don’t need to rely on overspecified prompts. They tailor real-world scenarios to whatever domain you’re focused on, and they make sure the data you train on rewards real understanding, not just instruction-following. Learn more at labelbox.com/dwarkesh* Mercury’s personal accounts let you add users, issue cards, and customize permissions. This is super useful for sharing finances with a partner, a roommate… or even an OpenClaw agent. And, if you’re already a Mercury Business user, your personal account is free! See terms and conditions below, and learn more at mercury.com/personal-bankingEligible Mercury Business users who apply for and maintain a Mercury Personal account may have their Mercury Personal subscription fee waived provided they remain a user on an active Mercury Business account in good standing. Standard Mercury Platform Subscription fees will apply if they no longer meet eligibility requirements, including but not limited to no longer being associated with an eligible Mercury Business account, or if the program is modified or terminated. Mercury may modify or discontinue this offering at any time and will provide notice as required by law. See Subscription Terms for full details.* To sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) - How cosplaying Ancient Rome led to the Renaissance(00:28:49) - How Florence’s weird republic worked(00:38:13) - How the Medicis took over Florence(00:58:12) - Why it was so hard for Gutenberg to make any money off the printing press(01:17:34) - Why the industrial revolution didn’t happen in Italy(01:23:02) - The Library of Alexandria isn’t where most ancient books were lost(01:41:21) - The Inquisition accidentally invented peer review This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Why Leonardo was a saboteur, Gutenberg went broke, and Florence was weird – Ada Palmer

Dario Amodei thinks we are just a few years away from AGI — or as he puts it, from having “a country of geniuses in a data center”. In this episode, we discuss what to make of the scaling hypothesis in the current RL regime, why task-specific RL might lead to generalization, and how AI will diffuse throughout the economy. We also dive into Anthropic’s revenue projections, compute commitments, path to profitability, and more.Watch on YouTube; read the transcript.Sponsors* Labelbox can get you the RL tasks and environments you need. Their massive network of subject-matter experts ensures realism across domains, and their in-house tooling lets them continuously tweak task difficulty to optimize learning. Reach out at labelbox.com/dwarkesh.* Jane Street sent me another puzzle… this time, they’ve trained backdoors into 3 different language models — they want you to find the triggers. Jane Street isn’t even sure this is possible, but they’ve set aside $50,000 for the best attempts and write-ups. They’re accepting submissions until April 1st at janestreet.com/dwarkesh.* Mercury’s personal accounts make it easy to share finances with a partner, a roommate… or OpenClaw. Last week, I wanted to try OpenClaw for myself, so I used Mercury to spin up a virtual debit card with a small spend limit, and then I let my agent loose. No matter your use case, apply at mercury.com/personal-banking.Timestamps(00:00:00) - What exactly are we scaling?(00:12:36) - Is diffusion cope?(00:29:42) - Is continual learning necessary?(00:46:20) - If AGI is imminent, why not buy more compute?(00:58:49) - How will AI labs actually make profit?(01:31:19) - Will regulations destroy the boons of AGI?(01:47:41) - Why can’t China and America both have a country of geniuses in a datacenter? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Dario Amodei — "We are near the end of the exponential"

In this episode, John and I got to do a real deep-dive with Elon. We discuss the economics of orbital data centers, the difficulties of scaling power on Earth, what it would take to manufacture humanoids at high-volume in America, xAI’s business and alignment plans, DOGE, and much more.Watch on YouTube; read the transcript.Sponsors* Mercury just started offering personal banking! I’m already banking with Mercury for business purposes, so getting to bank with them for my personal life makes everything so much simpler. Apply now at mercury.com/personal-banking* Jane Street sent me a new puzzle last week: they trained a neural net, shuffled all 96 layers, and asked me to put them back in order. I tried but… I didn’t quite nail it. If you’re curious, or if you think you can do better, you should take a stab at janestreet.com/dwarkesh* Labelbox can get you robotics and RL data at scale. Labelbox starts by helping you define your ideal data distribution, and then their massive Alignerr network collects frontier-grade data that you can use to train your models. Learn more at labelbox.com/dwarkeshTimestamps(00:00:00) - Orbital data centers(00:36:46) - Grok and alignment(00:59:56) - xAI’s business plan(01:17:21) - Optimus and humanoid manufacturing(01:30:22) - Does China win by default?(01:44:16) - Lessons from running SpaceX(02:20:08) - DOGE(02:38:28) - TeraFab This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Elon Musk — "In 36 months, the cheapest place to put AI will be space”

Adam Marblestone is CEO of Convergent Research. He’s had a very interesting past life: he was a research scientist at Google Deepmind on their neuroscience team and has worked on everything from brain-computer interfaces to quantum computing to nanotech and even formal mathematics.In this episode, we discuss how the brain learns so much from so little, what the AI field can learn from neuroscience, and the answer to Ilya’s question: how does the genome encode abstract reward functions? Turns out, they’re all the same question.Watch on YouTube; read the transcript.Sponsors* Gemini 3 Pro recently helped me run an experiment to test multi-agent scaling: basically, if you have a fixed budget of compute, what is the optimal way to split it up across agents? Gemini was my colleague throughout the process — honestly, I couldn’t have investigated this question without it. Try Gemini 3 Pro today gemini.google.com* Labelbox helps you train agents to do economically-valuable, real-world tasks. Labelbox’s network of subject-matter experts ensures you get hyper-realistic RL environments, and their custom tooling lets you generate the highest-quality training data possible from those environments. Learn more at labelbox.com/dwarkeshTo sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) – The brain’s secret sauce is the reward functions, not the architecture(00:22:20) – Amortized inference and what the genome actually stores(00:42:42) – Model-based vs model-free RL in the brain(00:50:31) – Is biological hardware a limitation or an advantage?(01:03:59) – Why a map of the human brain is important(01:23:28) – What value will automating math have?(01:38:18) – Architecture of the brainFurther readingIntro to Brain-Like-AGI Safety - Steven Byrnes’s theory of the learning vs steering subsystem; referenced throughout the episode.A Brief History of Intelligence - Great book by Max Bennett on connections between neuroscience and AIAdam’s blog, and Convergent Research’s blog on essential technologies.A Tutorial on Energy-Based Learning by Yann LeCunWhat Does It Mean to Understand a Neural Network? - Kording & LillicrapE11 Bio and their brain connectomics approachSam Gershman on what dopamine is doing in the brainGwern’s proposal on training models on the brain’s hidden states This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Adam Marblestone — AI is missing something fundamental about the brain

Read the essay here.Timestamps00:00:00 What are we scaling?00:03:11 The value of human labor00:05:04 Economic diffusion lag is cope00:06:34 Goal-post shifting is justified00:08:23 RL scaling00:09:18 Broadly deployed intelligence explosion This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Thoughts on AI progress (Dec 2025)

This is the final episode of the Sarah Paine lecture series, and it’s probably my favorite one. Sarah gives a “tour of the arguments” on what ultimately led to the Soviet Union’s collapse, diving into the role of the US, the Sino-Soviet border conflict, the oil bust, ethnic rebellions and even the Roman Catholic Church. As she points out, this is all particularly interesting as we find ourselves potentially at the beginning of another Cold War.As we wrap up this lecture series, I want to take a moment to thank Sarah for doing this with me. It has been such a pleasure.If you want more of her scholarship, I highly recommend checking out the books she’s written. You can find them here.Watch on YouTube; read the transcript.Sponsors* Labelbox can get you the training data you need, no matter the domain. Their Alignerr network includes the STEM PhDs and coding experts you’d expect, but it also has experienced cinematographers and talented voice actors to help train frontier video and audio models. Learn more at labelbox.com/dwarkesh.* Sardine doesn’t just assess customer risk for banking & retail. Their AI risk management platform is also extremely good at detecting fraudulent job applications, which I’ve found useful for my own hiring process. If you need help with hiring risk—or any other type of fraud prevention—go to sardine.ai/dwarkesh.* Gemini’s Nano Banana Pro helped us make many of the visuals in this episode. For example, we used it to turn dense tables into clear charts so that’d it be easier to quickly understand the trends that Sarah discusses. You can try Nano Banana Pro now in the Gemini app. Go to gemini.google.com.Timestamps(00:00:00) – Did Reagan single-handedly win the Cold War?(00:15:53) – Eastern Bloc uprisings & oil crisis(00:30:37) – Gorbachev’s mistakes(00:37:33) – German unification and NATO expansion(00:48:31) – The Gulf War and the Cold War endgame(00:56:10) – How central planning survived so long(01:14:46) – Sarah’s life in the USSR in 1988 This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Sarah Paine — Why Russia Lost the Cold War

Ilya & I discuss SSI’s strategy, the problems with pre-training, how to improve the generalization of AI models, and how to ensure AGI goes well.Watch on YouTube; read the transcript.Sponsors* Gemini 3 is the first model I’ve used that can find connections I haven’t anticipated. I recently wrote a blog post on RL’s information efficiency, and Gemini 3 helped me think it all through. It also generated the relevant charts and ran toy ML experiments for me with zero bugs. Try Gemini 3 today at gemini.google* Labelbox helped me create a tool to transcribe our episodes! I’ve struggled with transcription in the past because I don’t just want verbatim transcripts, I want transcripts reworded to read like essays. Labelbox helped me generate the exact data I needed for this. If you want to learn how Labelbox can help you (or if you want to try out the transcriber tool yourself), go to labelbox.com/dwarkesh* Sardine is an AI risk management platform that brings together thousands of device, behavior, and identity signals to help you assess a user’s risk of fraud & abuse. Sardine also offers a suite of agents to automate investigations so that as fraudsters use AI to scale their attacks, you can use AI to scale your defenses. Learn more at sardine.ai/dwarkeshTo sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) – Explaining model jaggedness(00:09:39) - Emotions and value functions(00:18:49) – What are we scaling?(00:25:13) – Why humans generalize better than models(00:35:45) – SSI’s plan to straight-shot superintelligence(00:46:47) – SSI’s model will learn from deployment(00:55:07) – How to think about powerful AGIs(01:18:13) – “We are squarely an age of research company”(01:20:23) – Self-play and multi-agent(01:32:42) – Research taste This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Ilya Sutskever — We're moving from the age of scaling to the age of research

As part of this interview, Satya Nadella gave Dylan Patel (founder of SemiAnalysis) and me an exclusive first-look at their brand-new Fairwater 2 datacenter.Microsoft is building multiple Fairwaters, each of which has hundreds of thousands of GB200s & GB300s. Between all these interconnected buildings, they’ll have over 2 GW of total capacity. Just to give a frame of reference, even a single one of these Fairwater buildings is more powerful than any other AI datacenter that currently exists.Satya then answered a bunch of questions about how Microsoft is preparing for AGI across all layers of the stack.Watch on YouTube; read the transcript.Sponsors* Labelbox produces high-quality data at massive scale, powering any capability you want your model to have. Whether you’re building a voice agent, a coding assistant, or a robotics model, Labelbox gets you the exact data you need, fast. Reach out at labelbox.com/dwarkesh* CodeRabbit automatically reviews and summarizes PRs so you can understand changes and catch bugs in half the time. This is helpful whether you’re coding solo, collaborating with agents, or leading a full team. To learn how CodeRabbit integrates directly into your workflow, go to coderabbit.aiTo sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) - Fairwater 2(00:03:20) - Business models for AGI(00:12:48) - Copilot(00:20:02) - Whose margins will expand most?(00:36:17) - MAI(00:47:47) - The hyperscale business(01:02:44) - In-house chip & OpenAI partnership(01:09:35) - The CAPEX explosion(01:15:07) - Will the world trust US companies to lead AI? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Satya Nadella — How Microsoft is preparing for AGI

In this lecture, military historian Sarah Paine explains how Russia—and specifically Stalin—completely derailed China’s rise, slowing them down for over a century.This lecture was particularly interesting to me because, in my opinion, the Chinese Civil War is 1 of the top 3 most important events of the 20th century. And to understand why it transpired as it did, you need to understand Stalin’s role in the whole thing.Watch on YouTube; read the transcript.SponsorsMercury helps you run your business better. It’s the banking platform we use for the podcast — we love that we can see our cash balance, AR, and AP all in one place. Join us (and over 200,000 other entrepreneurs) at mercury.comLabelbox scrutinizes public benchmarks at the single data-row level to probe what’s really being evaluated. Using this knowledge, they can generate custom training data for hill climbing existing benchmarks, or design new benchmarks from scratch. Learn more at labelbox.com/dwarkeshTo sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) – How Russia took advantage of China’s weakness(00:22:58) – After Stalin, China’s rise(00:33:52) – Russian imperialism(00:45:23) – China’s and Russia’s existential problems(01:04:55) – Q&A: Sino-Soviet Split(01:22:44) – Stalin’s lessons from WW2 This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Sarah Paine — How Russia sabotaged China's rise

The Andrej Karpathy episode.During this interview, Andrej explains why reinforcement learning is terrible (but everything else is much worse), why AGI will just blend into the previous ~2.5 centuries of 2% GDP growth, why self driving took so long to crack, and what he sees as the future of education.It was a pleasure chatting with him.Watch on YouTube; read the transcript.Sponsors* Labelbox helps you get data that is more detailed, more accurate, and higher signal than you could get by default, no matter your domain or training paradigm. Reach out today at labelbox.com/dwarkesh* Mercury helps you run your business better. It’s the banking platform we use for the podcast — we love that we can see our accounts, cash flows, AR, and AP all in one place. Apply online in minutes at mercury.com* Google’s Veo 3.1 update is a notable improvement to an already great model. Veo 3.1’s generations are more coherent and the audio is even higher-quality. If you have a Google AI Pro or Ultra plan, you can try it in Gemini today by visiting https://gemini.googleTimestamps(00:00:00) – AGI is still a decade away(00:29:45) – LLM cognitive deficits(00:40:05) – RL is terrible(00:49:38) – How do humans learn?(01:06:25) – AGI will blend into 2% GDP growth(01:17:36) – ASI(01:32:50) – Evolution of intelligence & culture(01:42:55) - Why self driving took so long(01:56:20) - Future of education This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Andrej Karpathy — AGI is still a decade away

Nick Lane has some pretty wild ideas about the evolution of life.He thinks early life was continuous with the spontaneous chemistry of undersea hydrothermal vents.Nick’s story may be wrong, but I find it remarkable that with just that starting point, you can explain so much about why life is the way that it is — the things you’re supposed to just take as givens in biology class:* Why are there two sexes? Why sex at all?* Why are bacteria so simple despite being around for 4 billion years? Why is there so much shared structure between all eukaryotic cells despite the enormous morphological variety between animals, plants, fungi, and protists?* Why did the endosymbiosis event that led to eukaryotes happen only once, and in the particular way that it did?* Why is all life powered by proton gradients? Why does all life on Earth share not only the Krebs Cycle, but even the intermediate molecules like Acetyl-CoA?His theory implies that early life is almost chemically inevitable (potentially blooming on hundreds of millions of planets in the Milky Way alone), and that the real bottleneck is the complex eukaryotic cell.Watch on YouTube; listen on Apple Podcasts or Spotify.Sponsors* Gemini in Sheets lets you turn messy text into structured data. We used it to classify all our episodes by type and topic, no manual tagging required. If you’re a Google Workspace user, you can get started today at docs.google.com/spreadsheets/* Labelbox has a massive network of domain experts (called Alignerrs) who help train AI models in a way that ensures they understand the world deeply, not superficially. These Alignerrs are true experts — one even tutored me in chemistry as I prepped for this episode. Learn more at labelbox.com/dwarkesh* Lighthouse helps frontier technology companies like Cursor and Physical Intelligence navigate the U.S. immigration system and hire top talent from around the world. Lighthouse handles everything, maximizing the probability of visa approval while minimizing the work you have to do. Learn more at lighthousehq.com/employersTo sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) – The singularity that unlocked complex life(00:08:26) – Early life continuous with Earth's geochemistry(00:23:36) – Eukaryotes are the great filter for intelligent life(00:42:16) – Mitochondria are the reason we have sex(01:08:12) – Are bioelectric fields linked to consciousness?Ref: 868329 This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Nick Lane – Life as we know it is chemically inevitable

I have a much better understanding of Sutton’s perspective now. I wanted to reflect on it a bit.(00:00:00) - The steelman(00:02:42) - TLDR of my current thoughts(00:03:22) - Imitation learning is continuous with and complementary to RL(00:08:26) - Continual learning(00:10:31) - Concluding thoughts This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Some thoughts on the Sutton interview

Richard Sutton is the father of reinforcement learning, winner of the 2024 Turing Award, and author of The Bitter Lesson. And he thinks LLMs are a dead end.After interviewing him, my steel man of Richard’s position is this: LLMs aren’t capable of learning on-the-job, so no matter how much we scale, we’ll need some new architecture to enable continual learning.And once we have it, we won’t need a special training phase — the agent will just learn on-the-fly, like all humans, and indeed, like all animals.This new paradigm will render our current approach with LLMs obsolete.In our interview, I did my best to represent the view that LLMs might function as the foundation on which experiential learning can happen… Some sparks flew.A big thanks to the Alberta Machine Intelligence Institute for inviting me up to Edmonton and for letting me use their studio and equipment.Enjoy!Watch on YouTube; listen on Apple Podcasts or Spotify.Sponsors* Labelbox makes it possible to train AI agents in hyperrealistic RL environments. With an experienced team of applied researchers and a massive network of subject-matter experts, Labelbox ensures your training reflects important, real-world nuance. Turn your demo projects into working systems at labelbox.com/dwarkesh* Gemini Deep Research is designed for thorough exploration of hard topics. For this episode, it helped me trace reinforcement learning from early policy gradients up to current-day methods, combining clear explanations with curated examples. Try it out yourself at gemini.google.com* Hudson River Trading doesn’t silo their teams. Instead, HRT researchers openly trade ideas and share strategy code in a mono-repo. This means you’re able to learn at incredible speed and your contributions have impact across the entire firm. Find open roles at hudsonrivertrading.com/dwarkeshTimestamps(00:00:00) – Are LLMs a dead end?(00:13:04) – Do humans do imitation learning?(00:23:10) – The Era of Experience(00:33:39) – Current architectures generalize poorly out of distribution(00:41:29) – Surprises in the AI field(00:46:41) – Will The Bitter Lesson still apply post AGI?(00:53:48) – Succession to AIs This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Richard Sutton – Father of RL thinks LLMs are a dead end

Sergey Levine, one of the world’s top robotics researchers and co-founder of Physical Intelligence, thinks we’re on the cusp of a “self-improvement flywheel” for general-purpose robots. His median estimate for when robots will be able to run households entirely autonomously? 2030.If Sergey’s right, the world 5 years from now will be an insanely different place than it is today. This conversation focuses on understanding how we get there: we dive into foundation models for robotics, and how we scale both the data and the hardware necessary to enable a full-blown robotics explosion.Watch on YouTube; listen on Apple Podcasts or Spotify.Sponsors* Labelbox provides high-quality robotics training data across a wide range of platforms and tasks. From simple object handling to complex workflows, Labelbox can get you the data you need to scale your robotics research. Learn more at labelbox.com/dwarkesh* Hudson River Trading uses cutting-edge ML and terabytes of historical market data to predict future prices. I got to try my hand at this fascinating prediction problem with help from one of HRT’s senior researchers. If you’re curious about how it all works, go to hudson-trading.com/dwarkesh* Gemini 2.5 Flash Image (aka nano banana) isn’t just for generating fun images — it’s also a powerful tool for restoring old photos and digitizing documents. Test it yourself in the Gemini App or in Google’s AI Studio: ai.studio/bananaTo sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) – Timeline to widely deployed autonomous robots(00:17:25) – Why robotics will scale faster than self-driving cars(00:27:28) – How vision-language-action models work(00:45:37) – Changes needed for brainlike efficiency in robots(00:57:59) – Learning from simulation(01:09:18) – How much will robots speed up AI buildouts?(01:18:01) – If hardware’s the bottleneck, does China win by default? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Fully autonomous robots are much closer than you think – Sergey Levine

In this lecture, military historian Sarah Paine explains how Britain used sea control, peripheral campaigns, and alliances to defeat Nazi Germany during WWII. She then applies this framework to today, arguing that Russia and China are similarly constrained by their geography, making them vulnerable in any conflict with maritime powers (like the U.S. and its allies).Watch on YouTube; listen on Apple Podcasts or Spotify.Sponsors* Labelbox partners with researchers to scope, generate, and deliver the exact data frontier models need, no matter the domain. Whether that’s multi-turn audio, SOTA robotics data, advanced STEM problem sets, or even novel RL environments, Labelbox delivers high-quality data, fast. Learn more at labelbox.com/dwarkesh* Warp is the best interface I’ve found for coding with agents. It makes building custom tools easy: Warp’s UI helps you understand agent behavior and its in-line text editor is great for making tweaks. You can try Warp for free, or, for a limited time, use code DWARKESH to get Warp’s Pro Plan for only $5. Go to warp.dev/dwarkeshTo sponsor a future episode, visit dwarkesh.com/advertise.Timestamps00:00:00 – How WW1 shaped WW200:15:10 – Hitler and Churchill’s battle to command the Atlantic00:30:10 – Peripheral theaters leading up to Normandy00:37:13 – The Eastern front00:48:04 – Russia’s & China’s geographic prisons01:00:28 – Hitler’s blunders & America’s industrial might01:15:03 – Bismarck’s limited wars vs Hitler’s total war This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

How Hitler almost starved Britain – Sarah Paine

Jacob Kimmel thinks he can find the transcription factors to reverse aging. We do a deep dive on why this might be plausible and why evolution hasn’t optimized for longevity. We also talk about why drug discovery has been getting exponentially harder, and what a new platform for biological understanding to speed up progress would look like. As a bonus, we get into the nitty gritty of gene delivery and Jacob’s controversial takes on CAR-T cells. For full disclosure, I am an angel investor in NewLimit. This did not impact my decision to interview Jacob, nor the questions I asked him.Watch on YouTube; listen on Apple Podcasts or Spotify.SPONSORS* Hudson River Trading uses deep learning to tackle one of the world's most complex systems: global capital allocation. They have a massive in-house GPU cluster, and they’re constantly adding new racks of B200s to ensure their researchers are never constrained by compute. Explore opportunities at hudsonrivertrading.com/dwarkesh\* Google’s Gemini CLI turns ideas into working applications FAST, no coding required. It built a complete podcast post-production tool in 10 minutes, including fully functional backend logic, and the entire build used less than 10% of Gemini’s session context. Check it out on Github now!* To sponsor a future episode, visit dwarkesh.com/advertise.TIMESTAMPS(00:00:00) – Three reasons evolution didn’t optimize for longevity(00:12:07) – Why didn't humans evolve their own antibiotics?(00:25:26) – De-aging cells via epigenetic reprogramming(00:44:43) – Viral vectors and other delivery mechanisms(01:06:22) – Synthetic transcription factors(01:09:31) – Can virtual cells break Eroom’s Law?(01:31:32) – Economic models for pharma This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Evolution designed us to die fast; we can change that — Jacob Kimmel

How will we feed the 100s of GWs of extra energy demand that AI will create over the coming decade? On this episode, Casey Handmer (Caltech PhD, former NASA JPL, founder & CEO of Terraform Industries) walks me through how we can pull it off, and why he thinks a major part of this energy singularity will be powered by solar. His views are contrarian, but he came armed to defend them.Watch on YouTube; listen on Apple Podcasts or Spotify.SPONSORS- Lighthouse helps frontier technology companies like Cursor and Physical Intelligence navigate the U.S. immigration system and hire top talent from around the world. Lighthouse handles everything for you, maximizing the probability of visa approval while minimizing the work you have to do. Learn more at lighthousehq.com/employers- To sponsor a future episode, visit dwarkesh.com/advertise.TIMESTAMPS(00:00:00) – Why doesn’t China win by default?(00:08:28) – Why hyperscalers choose natural gas over solar(00:18:01) – Solar's astonishing learning rates(00:27:02) – How to build 50,000 acre solar-powered data centers(00:40:24) – Environmental regulations blocking clean energy(00:44:04) – Batteries replacing the grid(00:49:14) – GDP is broken, AGI's true value must be measured in total energy use(00:58:45) – Silicon wafers in space with one mind each This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

China is killing the US on energy. Does that mean they’ll win AGI? — Casey Handmer

A deep dive with Lewis Bollard, who leads Open Philanthropy’s strategy for Farmed Animal Welfare, on the surprising economics of the meat industry.Why is factory farming so efficient? How can we make the lives of the 23+ billion animals living on factory farms more bearable? How far off are the moonshots (e.g., brainless chickens, cultivated meats, etc.) to end this mass suffering? And why does the meat industry have such a surprising amount of political influence?For decades, innovation in the meat industry has actually made the conditions for animals worse. Can the next few decades of tech reverse this pattern?Watch on YouTube; listen on Apple Podcasts or Spotify.Donation match fundraiserThe welfare of animals on factory farms is so systemically neglected that just $1 can help avert 10 years of animal suffering.After learning more about the outsized opportunities to help, I decided to give $250,000 as a donation match to farmkind.giving/dwarkesh. FarmKind directs your contributions to the most effective charities in this area.Please consider contributing, even if it’s a small amount. Together, we can double each other's impact and give a total of $500,000.Bluntly, there are some listeners who are in a position to give much more. Given how neglected this topic is, one such person could singlehandedly change the game for 10s of billions of animals. If you’re considering donating $50k or more, please reach out directly to Lewis and his team by emailing andres@openphilanthropy.org.Timestamps(00:00:00) – The astonishing efficiency of factory farming(00:07:18) – It was a mistake making this about diet(00:09:54) – Tech that’s sparing 100s of millions of animals/year(00:16:16) – Brainless chickens and higher welfare breeds(00:28:21) – $1 can prevent 10 years of animal suffering(00:37:26) – Situation in China and the developing world(00:41:41) – How the meat lobby got a lock on Congress(00:53:23) – Business structure of the meat industry(00:57:42) – Corporate campaigns are underrated This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Artificial meat is harder than artificial intelligence — Lewis Bollard

After my last lecture series with Sarah Paine ended, I still had so many questions. I knew we’d only scratched the surface of Sarah’s scholarship, so I immediately invited her back for another series: she graciously agreed, and we’ll be releasing the results online over the coming weeks and months!This first lecture is focused on the balance of power in East Asia at the turn of the 20th century. Specifically, how did Japan (population 47M) defeat China (400M) and Russia (130M) to become Asia's dominant power?For me, the most interesting thing was that Japan's surprise attack on Port Arthur at the beginning of the Russo-Japanese War (1904) helps us understand why Japan might have thought Pearl Harbor would work.Watch on YouTube; listen on Apple Podcasts or Spotify.Sponsors* Google’s Veo 3 helps us visualize the hypothetical scenarios that often come up during our interviews. Veo’s ability to generate both video and audio—all with incredible realism—makes it perfect for bringing our content to life. If you have a Google AI Pro or Ultra plan, you can try it in Gemini today by visiting gemini.google.* Hudson River Trading is one of the world's top quantitative trading firms. Responsible for around 15% of all U.S. equities trading volume, HRT powers their trades with cutting-edge deep learning models. Their in-house AI team does fundamental ML research and then applies it to some of the most competitive markets in the world. If you’re interested in joining them, you can learn more at hudsonrivertrading.com/dwarkesh.To sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) – Japan’s Meiji reforms(00:14:44) – Trans-Siberian railway & Japan’s 3-year window for empire(00:29:58) – The most important battle in the Russo-Japanese war(00:48:38) – China’s implosion: imperialism, civil wars, and opium(00:59:31) – Was Russia on track to dominate Asia?(01:14:20) – Pearl Harbor (1941) vs surprise attack of Port Arthur (1904)(01:34:03) – Why big countries still lose wars(01:46:56) – Grand strategy for small countries This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Sarah Paine — How Imperial Japan defeated Tsarist Russia & Qing China

The Stephen Kotkin episode. Kotkin is arguably the world’s foremost expert on Joseph Stalin and has written a massive 2-volume biography on him (with a 3rd volume in the works).No other individual had more of a profound impact on the 20th century than Stalin. He held the power of life and death over every single person across 11 time zones, and he killed tens of millions of people, utterly consumed by an ideology aimed at building paradise on Earth.And, he was one half of the biggest and most consequential military confrontation in history (even if Hitler didn’t prove to be his match).Watch on YouTube; listen on Apple Podcasts or Spotify.Sponsors* Lighthouse is THE fastest immigration solution for the technology industry. All they need is your resume or LinkedIn profile to tell you which visas you’re most eligible for, and they’ll send you this eligibility document for free, no commitment required. Get started today at https://www.lighthousehq.com/ref/Dwarkesh.To sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) – Was the tsarist regime the lesser of 2 evils?(00:23:45) – The peasants brought Lenin to power, then he enslaved them(00:37:38) – Why did so many go along with enforced famine and the Great Terror?(01:02:26) – Today’s leftist civil war(01:13:01) – Doesn’t CCP deserve credit for China's growth?(01:35:13) – Why didn't somebody just kill Stalin?(01:52:45) – Overcoming the pathologies of communism with tech: USSR vs China This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Stephen Kotkin — How Stalin became the most powerful dictator in history

I’ve had a lot of discussions on my podcast where we haggle out timelines to AGI. Some guests think it’s 20 years away - others 2 years. Here’s an audio version of where my thoughts stand as of June 2025. If you want to read the original post, you can check it out here. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Why I don’t think AGI is right around the corner

George Church is the godfather of modern synthetic biology and has been involved with basically every major biotech breakthrough in the last few decades.Professor Church thinks that these improvements (e.g., orders of magnitude decrease in sequencing & synthesis costs, precise gene editing tools like CRISPR, AlphaFold-type AIs, & the ability to conduct massively parallel multiplex experiments) have put us on the verge of some massive payoffs: de-aging, de-extinction, biobots that combine the best of human and natural engineering, and (unfortunately) weaponized mirror life.Watch on YouTube; listen on Apple Podcasts or Spotify.Sponsors* WorkOS Radar ensures your product is ready for AI agents. Radar is an anti-fraud solution that categorizes different types of automated traffic, blocking harmful bots while allowing helpful agents. Future-proof your roadmap today at workos.com/radar.* Scale is building the infrastructure for smarter, safer AI. In addition to their Data Foundry, they recently released Scale Evaluation, a tool that diagnoses model limitations. Learn how Scale can help you push the frontier at scale.com/dwarkesh.* Gemini 2.5 Pro was invaluable during our prep for this episode: it perfectly explained complex biology and helped us understand the most important papers. Gemini’s recently improved structure and style also made using it surprisingly enjoyable. Start building with it today at https://aistudio.google.comTo sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(0:00:00) – Aging solved by 2050(0:07:37) – Finding the master switch for any trait(0:19:50) – Weaponized mirror life(0:30:40) – Why hasn’t sequencing/synthesis led to biotech revolution?(0:50:26) – Impact of AGI on biology research progress(1:00:35) – Biobots that use the best of biological and human engineering(1:05:09) – Odds of life in universe(1:09:57) – Is DNA the ultimate data storage?(1:13:55) – Curing rare diseases with genetic counseling(1:22:23) – NIH & NSF budget cuts(1:25:26) – How one lab spawned 100 biotech companies This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

A billion years of evolution in a single afternoon — George Church

Arthur Kroeber is a leading researcher on Chinese tech and macro, a founding partner at Gavekal Dragonomics, and author of "China's Economy: What Everyone Needs to Know." It's the most useful, detailed resource I've found of how China actually works.On this episode, we discuss how China achieved high-tech manufacturing dominance, and where they'll go from here. By Arthur’s account, the Chinese government is like a giant VC fund: they decide on key priorities and then spend hundreds of billions of dollars subsidizing ruthless competition at the local level. They are willing to lose huge amounts of money for a few of their bets to pay off: at China’s scale, effectiveness matters more than efficiency.There's also a growing bipartisan consensus that we need to combat China's rise. This doesn’t make much sense to me. China is a big, powerful country at the frontier in many fields, and its economy is intricately tied in with our own. Being instinctively adversarial is both unsustainable and risky. Arthur and I discuss how we can create a productive, mutually beneficial version of this relationship.Watch on YouTube; listen on Apple Podcasts or Spotify.Sponsors* Scale is building the infrastructure for smarter, safer AI. In addition to their Data Foundry, they recently released Scale Evaluation, a tool that diagnoses model limitations. Learn how Scale can help you push the frontier at scale.com/dwarkesh.* WorkOS Radar ensures your product’s free trials go to actual users. Radar uses 80+ signals to distinguish malicious bots from real people, eliminating costly free-tier abuse. See why companies like Cursor, Perplexity, and OpenAI use Radar by visiting workos.com/radar.* Lighthouse is THE fastest immigration solution for the technology industry. They help you understand your options and navigate applications for expert visas like the O-1A and EB-1A. Explore which visa is right for you at https://www.lighthousehq.com/ref/Dwarkesh.To sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) – We should reconcile with China(00:21:21) – BYD, Tesla, & Chinese EV industry(00:36:05) – Will China have a Japan-style financial crisis?(00:44:39) – Local debt situation is manageable(00:57:28) – If CCP is so competent, why isn’t China richer?(01:05:08) – How China keeps tech under control(01:33:45) – Does China win AI?(01:43:34) – Communication with China key for AI safety(02:10:08) – What foreigners get wrong about China(02:17:32) – China-US relationship future This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Why China's manufacturing economy is dominating — Arthur Kroeber

Ken Rogoff is the former chief economist of the IMF, a professor of Economics at Harvard, and author of the newly released Our Dollar, Your Problem and This Time is Different.On this episode, Ken predicts that, within the next decade, the US will have a debt-induced inflation crisis, but not a Japan-type financial crisis (the latter is much worse, and can make a country poorer for generations).Ken also explains how China is trapped: in order to solve their current problems, they’ll keep leaning on financial repression and state-directed investment, which only makes their situation worse.We also discuss the erosion of dollar dominance, why there will be a rebalancing toward foreign equities, how AGI will impact the deficit and interest rate, and much more!Watch on YouTube; listen on Apple Podcasts or Spotify.Sponsors* WorkOS gives your product all the features that enterprise customers need, without derailing your roadmap. Skip months of engineering effort and start selling to enterprises today at workos.com.* Scale is building the infrastructure for smarter, safer AI. In addition to their Data Foundry, they recently released Scale Evaluation, a tool that diagnoses model limitations. Learn how Scale can help you push the frontier at scale.com/dwarkesh.* Gemini Live API lets you have natural, real-time, interactions with Gemini. You can talk to it like you were talking to another person, stream video to show it your surroundings, and share screen to give it context. Try it now by clicking the “Stream” tab on ai.dev.To sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) – China is stagnating(00:25:46) – How the US broke Japan's economy(00:37:06) – America's inflation crisis is coming(01:02:20) – Will AGI solve the US deficit?(01:07:11) – Why interest rates will go up(01:10:55) – US equities will underperform(01:22:24) – The erosion of dollar dominance This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

"China is digging out of a crisis. And America’s luck is wearing thin." — Ken Rogoff

On this episode, I chat with Victor Shih about all things China. We discuss China’s massive local debt crisis, the CCP’s views on AI, what happens after Xi, and more.Victor Shih is an expert on the Chinese political system, as well as their banking and fiscal policies, and he has amassed more biographical data on the Chinese elite than anyone else in the world. He teaches at UC San Diego, where he also directs the 21st Century China Center.Watch on YouTube; listen on Apple Podcasts or Spotify.Sponsors* Scale is building the infrastructure for smarter, safer AI. In addition to their Data Foundry, they just released Scale Evaluation, a tool that diagnoses model limitations. Learn how Scale can help you push the frontier at scale.com/dwarkesh.* WorkOS is how top AI companies ship critical enterprise features without burning months of engineering time. If you need features like SSO, audit logs, or user provisioning, head to workos.com.To sponsor a future episode, visit dwarkesh.com/advertise.Timestamps(00:00:00) – Is China more decentralized than the US?(00:03:16) – How the Politburo Standing Committee makes decisions(00:21:07) – Xi’s right hand man in charge of AGI(00:35:37) – DeepSeek was trained to track CCP policy(00:45:35) – Local government debt crisis(00:50:00) – BYD, CATL, & financial repression(00:58:12) – How corruption leads to overbuilding(01:10:46) – Probability of Taiwan invasion(01:18:56) – Succession after Xi(01:25:10) – Future growth forecasts This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Xi Jinping’s paranoid approach to AGI, debt crisis, & Politburo politics — Victor Shih

New episode with my good friends Sholto Douglas & Trenton Bricken. Sholto focuses on scaling RL and Trenton researches mechanistic interpretability, both at Anthropic.We talk through what’s changed in the last year of AI research; the new RL regime and how far it can scale; how to trace a model’s thoughts; and how countries, workers, and students should prepare for AGI.See you next year for v3. Here’s last year’s episode, btw. Enjoy!Watch on YouTube; listen on Apple Podcasts or Spotify.----------SPONSORS* WorkOS ensures that AI companies like OpenAI and Anthropic don't have to spend engineering time building enterprise features like access controls or SSO. It’s not that they don't need these features; it's just that WorkOS gives them battle-tested APIs that they can use for auth, provisioning, and more. Start building today at workos.com.* Scale is building the infrastructure for safer, smarter AI. Scale’s Data Foundry gives major AI labs access to high-quality data to fuel post-training, while their public leaderboards help assess model capabilities. They also just released Scale Evaluation, a new tool that diagnoses model limitations. If you’re an AI researcher or engineer, learn how Scale can help you push the frontier at scale.com/dwarkesh.* Lighthouse is THE fastest immigration solution for the technology industry. They specialize in expert visas like the O-1A and EB-1A, and they’ve already helped companies like Cursor, Notion, and Replit navigate U.S. immigration. Explore which visa is right for you at lighthousehq.com/ref/Dwarkesh.To sponsor a future episode, visit dwarkesh.com/advertise.----------TIMESTAMPS(00:00:00) – How far can RL scale?(00:16:27) – Is continual learning a key bottleneck?(00:31:59) – Model self-awareness(00:50:32) – Taste and slop(01:00:51) – How soon to fully autonomous agents?(01:15:17) – Neuralese(01:18:55) – Inference compute will bottleneck AGI(01:23:01) – DeepSeek algorithmic improvements(01:37:42) – Why are LLMs ‘baby AGI’ but not AlphaZero?(01:45:38) – Mech interp(01:56:15) – How countries should prepare for AGI(02:10:26) – Automating white collar work(02:15:35) – Advice for students This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Is RL + LLMs enough for AGI? — Sholto Douglas & Trenton Bricken

Based on my essay about AI firms.Huge thanks to Petr and his team for bringing this to life!Watch on YouTube.Thanks to Google for sponsoring. We used their Veo 2 model to make this entire video—it generated everything from the photorealistic humans to the claymation octopuses. If you’re a Gemini Advanced user, you can try Veo 2 now in the Gemini app. Just select Veo 2 in the dropdown, and type your video idea in the prompt bar. Get started today by going to gemini.google.com.To sponsor a future episode, visit dwarkesh.com/advertise. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

What will automated firms look like?

Zuck on:* Llama 4, benchmark gaming* Intelligence explosion, business models for AGI* DeepSeek/China, export controls, & Trump* Orion glasses, AI relationships, and preventing reward-hacking from our tech.Watch on Youtube; listen on Apple Podcasts and Spotify.----------SPONSORS* Scale is building the infrastructure for safer, smarter AI. Scale’s Data Foundry gives major AI labs access to high-quality data to fuel post-training, while their public leaderboards help assess model capabilities. They also just released Scale Evaluation, a new tool that diagnoses model limitations. If you’re an AI researcher or engineer, learn how Scale can help you push the frontier at scale.com/dwarkesh.* WorkOS Radar protects your product against bots, fraud, and abuse. Radar uses 80+ signals to identify and block common threats and harmful behavior. Join companies like Cursor, Perplexity, and OpenAI that have eliminated costly free-tier abuse by visiting workos.com/radar.* Lambda is THE cloud for AI developers, with over 50,000 NVIDIA GPUs ready to go for startups, enterprises, and hyperscalers. By focusing exclusively on AI, Lambda provides cost-effective compute supported by true experts, including a serverless API serving top open-source models like Llama 4 or DeepSeek V3-0324 without rate limits, and available for a free trial at lambda.ai/dwarkesh.To sponsor a future episode, visit dwarkesh.com/p/advertise.----------TIMESTAMPS(00:00:00) – How Llama 4 compares to other models(00:11:34) – Intelligence explosion(00:26:36) – AI friends, therapists & girlfriends(00:35:10) – DeepSeek & China(00:39:49) – Open source AI(00:54:15) – Monetizing AGI(00:58:32) – The role of a CEO(01:02:04) – Is big tech aligning with Trump?(01:07:10) – 100x productivity This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Mark Zuckerberg — AI will write most Meta code in 18 months

800 years before the Black Death, the very same bacteria ravaged Rome, killing 60%+ of the population in many areas.Also, back-to-back volcanic eruptions caused a mini Ice Age, leaving Rome devastated by famine and disease.I chatted with historian Kyle Harper about this and much else:* Rome as a massive slave society* Why humans are more disease-prone than other animals* How agriculture made us physically smaller (Caesar at 5'5" was considered tall)Watch on Youtube; listen on Apple Podcasts or Spotify.----------SPONSORS* WorkOS makes it easy to become enterprise-ready. They have APIs for all the most common enterprise requirements—things like authentication, permissions, and encryption—so you can quickly plug them in and get back to building your core product. If you want to make your product enterprise-ready, join companies like Cursor, Perplexity and OpenAI, and head to workos.com.* Scale’s Data Foundry gives major AI labs access to high-quality data to fuel post-training, including advanced reasoning capabilities. If you’re an AI researcher or engineer, learn how Scale’s Data Foundry and research lab, SEAL, can help you go beyond the current frontier of capabilities at scale.com/dwarkeshTo sponsor a future episode, visit dwarkesh.com/advertise.----------KYLE'S BOOKS* The Fate of Rome: Climate, Disease, and the End of an Empire* Plagues upon the Earth: Disease and the Course of Human History* Slavery in the Late Roman World, AD 275-425----------TIMESTAMPS(00:00:00) - Plague's impact on Rome's collapse(00:06:24) - Rome's little Ice Age(00:11:51) - Why did progress stall in Rome's Golden Age?(00:23:55) - Slavery in Rome(00:36:22) - Was agriculture a mistake?(00:47:42) - Disease's impact on cognitive function(00:59:46) - Plague in India and Central Asia(01:05:16) - The next pandemic(01:16:48) - How Kyle uses LLMs(01:18:51) - De-extinction of lost species This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Why Rome actually fell: plagues, slavery, & ice age — Kyle Harper

Ege Erdil and Tamay Besiroglu have 2045+ timelines, think the whole "alignment" framing is wrong, don't think an intelligence explosion is plausible, but are convinced we'll see explosive economic growth (economy literally doubling every year or two).This discussion offers a totally different scenario than my recent interview with Scott and Daniel.Ege and Tamay are the co-founders of Mechanize (disclosure - I’m an angel investor), a startup dedicated to fully automating work. Before founding Mechanize, Ege and Tamay worked on AI forecasts at Epoch AI. Watch on Youtube; listen on Apple Podcasts or Spotify.----------Sponsors* WorkOS makes it easy to become enterprise-ready. With simple APIs for essential enterprise features like SSO and SCIM, WorkOS helps companies like Vercel, Plaid, and OpenAI meet the requirements of their biggest customers. To learn more about how they can help you do the same, visit workos.com* Scale’s Data Foundry gives major AI labs access to high-quality data to fuel post-training, including advanced reasoning capabilities. If you’re an AI researcher or engineer, learn about how Scale’s Data Foundry and research lab, SEAL, can help you go beyond the current frontier at scale.com/dwarkesh* Google's Gemini Pro 2.5 is the model we use the most at Dwarkesh Podcast: it helps us generate transcripts, identify interesting clips, and code up new tools. If you want to try it for yourself, it's now available in Preview with higher rate limits! Start building with it today at aistudio.google.com.----------Timestamps(00:00:00) - AGI will take another 3 decades(00:22:27) - Even reasoning models lack animal intelligence (00:45:04) - Intelligence explosion(01:00:57) - Ege & Tamay’s story(01:06:24) - Explosive economic growth(01:33:00) - Will there be a separate AI economy?(01:47:08) - Can we predictably influence the future?(02:19:48) - Arms race dynamic(02:29:48) - Is superintelligence a real thing?(02:35:45) - Reasons not to expect explosive growth(02:49:00) - Fully automated firms(02:54:43) - Will central planning work after AGI?(02:58:20) - Career advice This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

AGI is still 30 years away — Ege Erdil & Tamay Besiroglu

Scott and Daniel break down every month from now until the 2027 intelligence explosion.Scott Alexander is author of the highly influential blogs Slate Star Codex and Astral Codex Ten. Daniel Kokotajlo resigned from OpenAI in 2024, rejecting a non-disparagement clause and risking millions in equity to speak out about AI safety.We discuss misaligned hive minds, Xi and Trump waking up, and automated Ilyas researching AI progress.I came in skeptical, but I learned a tremendous amount by bouncing my objections off of them. I highly recommend checking out their new scenario planning document, AI 2027Watch on Youtube; listen on Apple Podcasts or Spotify.----------Sponsors* WorkOS helps today’s top AI companies get enterprise-ready. OpenAI, Cursor, Perplexity, Anthropic and hundreds more use WorkOS to quickly integrate features required by enterprise buyers. To learn more about how you can make the leap to enterprise, visit workos.com* Jane Street likes to know what's going on inside the neural nets they use. They just released a black-box challenge for Dwarkesh listeners, and I had a blast trying it out. See if you have the skills to crack it at janestreet.com/dwarkesh* Scale’s Data Foundry gives major AI labs access to high-quality data to fuel post-training, including advanced reasoning capabilities. If you’re an AI researcher or engineer, learn about how Scale’s Data Foundry and research lab, SEAL, can help you go beyond the current frontier at scale.com/dwarkeshTo sponsor a future episode, visit dwarkesh.com/advertise.----------Timestamps(00:00:00) - AI 2027(00:06:56) - Forecasting 2025 and 2026(00:14:41) - Why LLMs aren't making discoveries(00:24:33) - Debating intelligence explosion(00:49:45) - Can superintelligence actually transform science?(01:16:54) - Cultural evolution vs superintelligence(01:24:05) - Mid-2027 branch point(01:32:30) - Race with China(01:44:47) - Nationalization vs private anarchy(02:03:22) - Misalignment(02:14:52) - UBI, AI advisors, & human future(02:23:00) - Factory farming for digital minds(02:26:52) - Daniel leaving OpenAI(02:35:15) - Scott's blogging advice This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

AI 2027: month-by-month model of intelligence explosion — Scott Alexander & Daniel Kokotajlo

I recorded an AMA! I had a blast chatting with my friends Trenton Bricken and Sholto Douglas. We discussed my new book, career advice given AGI, how I pick guests, how I research for the show, and some other nonsense.My book, “The Scaling Era: An Oral History of AI, 2019-2025” is available in digital format now. Preorders for the print version are also open!Watch on YouTube; listen on Apple Podcasts or Spotify.Timestamps(0:00:00) - Book launch announcement(0:04:57) - AI models not making connections across fields(0:10:52) - Career advice given AGI(0:15:20) - Guest selection criteria(0:17:19) - Choosing to pursue the podcast long-term(0:25:12) - Reading habits(0:31:10) - Beard deepdive(0:33:02) - Who is best suited for running an AI lab?(0:35:16) - Preparing for fast AGI timelines(0:40:50) - Growing the podcast This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

AMA: career advice given AGI, how I research ft. Sholto & Trenton

Humans have not succeeded because of our raw intelligence.Marooned European explorers regularly starved to death in areas where foragers thrived for 1000s of years.I’ve always found this cultural evolution deeply mysterious.How do you discover the 10 steps for processing cassava so it won’t give you cyanide poisoning simply by trial and error?Has the human brain declined in size over the last 10,000 years because we outsourced cultural evolution to a larger collective brain?The most interesting part of the podcast is Henrich’s explanation of how the Catholic Church unintentionally instigated the Industrial Revolution through the dismantling of intensive kinship systems in medieval Europe.Watch on Youtube; listen on Apple Podcasts or Spotify.----------SponsorsScale partners with major AI labs like Meta, Google Deepmind, and OpenAI. Through Scale’s Data Foundry, labs get access to high-quality data to fuel post-training, including advanced reasoning capabilities. If you’re an AI researcher or engineer, learn about how Scale’s Data Foundry and research lab, SEAL, can help you go beyond the current frontier at scale.com/dwarkesh.To sponsor a future episode, visit dwarkesh.com/p/advertise.----------Joseph’s booksThe WEIRDest People in the WorldThe Secret of Our Success----------Timestamps(0:00:00) - Humans didn’t succeed because of raw IQ(0:09:27) - How cultural evolution works(0:20:48) - Why is human brain size declining?(0:32:00) - Will AGI have superhuman cultural learning?(0:42:34) - Why Industrial Revolution happened in Europe(0:55:30) - Why China, Rome, India got left behind(1:21:09) - Loss of cultural variance in modern world(1:31:20) - Is individual genius real?(1:43:49) - IQ and collective brains This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Joseph Henrich — Humans defeated smarter species with cultural evolution

I’m so excited with how this visualization of Notes on China turned out. Petr, thank you for such beautiful watercolor artwork. More to come!Watch on YouTube.----------Timestamps(0:00:00) - Intro(0:00:32) - Scale(0:05:50) - Vibes(0:11:14) - Youngsters(0:14:27) - Tech & AI(0:15:47) - Hearts & Minds(0:17:07) - On Travel This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Notes on China

Satya Nadella on: Why he doesn’t believe in AGI but does believe in 10% economic growth; Microsoft’s new topological qubit breakthrough and gaming world models;Whether Office commoditizes LLMs or the other way around. Watch on Youtube; listen on Apple Podcasts or Spotify.----------SponsorsScale partners with major AI labs like Meta, Google Deepmind, and OpenAI. Through Scale’s Data Foundry, labs get access to high-quality data to fuel post-training, including advanced reasoning capabilities. If you’re an AI researcher or engineer, learn about how Scale’s Data Foundry and research lab, SEAL, can help you go beyond the current frontier at scale.com/dwarkeshLinear's project management tools have become the default choice for product teams at companies like Ramp, CashApp, OpenAI, and Scale. These teams use Linear so they can stay close to their products and move fast. If you’re curious why so many companies are making the switch, visit linear.app/dwarkeshTo sponsor a future episode, visit dwarkeshpatel.com/p/advertise.----------Timestamps(0:00:00) - Intro(0:05:04) - AI won't be winner-take-all(0:15:18) - World economy growing by 10%(0:21:39) - Decreasing price of intelligence(0:30:19) - Quantum breakthrough(0:42:51) - How Muse will change gaming(0:49:51) - Legal barriers to AI(0:55:46) - Getting AGI safety right(1:04:59) - 34 years at Microsoft(1:10:46) - Does Satya Nadella believe in AGI? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Satya Nadella — Microsoft’s AGI plan & quantum breakthrough

This week I welcome on the show two of the most important technologists ever, in any field.Jeff Dean is Google's Chief Scientist, and through 25 years at the company, has worked on basically the most transformative systems in modern computing: from MapReduce, BigTable, Tensorflow, AlphaChip, to Gemini.Noam Shazeer invented or co-invented all the main architectures and techniques that are used for modern LLMs: from the Transformer itself, to Mixture of Experts, to Mesh Tensorflow, to Gemini and many other things.We talk about their 25 years at Google, going from PageRank to MapReduce to the Transformer to MoEs to AlphaChip – and maybe soon to ASI.My favorite part was Jeff's vision for Pathways, Google’s grand plan for a mutually-reinforcing loop of hardware and algorithmic design and for going past autoregression. That culminates in us imagining *all* of Google-the-company, going through one huge MoE model.And Noam just bites every bullet: 100x world GDP soon; let’s get a million automated researchers running in the Google datacenter; living to see the year 3000.Watch on Youtube; listen on Apple Podcasts or Spotify.SponsorsScale partners with major AI labs like Meta, Google Deepmind, and OpenAI. Through Scale’s Data Foundry, labs get access to high-quality data to fuel post-training, including advanced reasoning capabilities. If you’re an AI researcher or engineer, learn about how Scale’s Data Foundry and research lab, SEAL, can help you go beyond the current frontier at scale.com/dwarkeshCurious how Jane Street teaches their new traders? They use Figgie, a rapid-fire card game that simulates the most exciting parts of markets and trading. It’s become so popular that Jane Street hosts an inter-office Figgie championship every year. Download from the app store or play on your desktop at figgie.comMeter wants to radically improve the digital world we take for granted. They’re developing a foundation model that automates network management end-to-end. To do this, they just announced a long-term partnership with Microsoft for tens of thousands of GPUs, and they’re recruiting a world class AI research team. To learn more, go to meter.com/dwarkeshTo sponsor a future episode, visit dwarkeshpatel.com/p/advertiseTimestamps00:00:00 - Intro00:02:44 - Joining Google in 199900:05:36 - Future of Moore's Law00:10:21 - Future TPUs00:13:13 - Jeff’s undergrad thesis: parallel backprop00:15:10 - LLMs in 200700:23:07 - “Holy s**t” moments00:29:46 - AI fulfills Google’s original mission00:34:19 - Doing Search in-context00:38:32 - The internal coding model00:39:49 - What will 2027 models do?00:46:00 - A new architecture every day?00:49:21 - Automated chip design and intelligence explosion00:57:31 - Future of inference scaling01:03:56 - Already doing multi-datacenter runs01:22:33 - Debugging at scale01:26:05 - Fast takeoff and superalignment01:34:40 - A million evil Jeff Deans01:38:16 - Fun times at Google01:41:50 - World compute demand in 203001:48:21 - Getting back to modularity01:59:13 - Keeping a giga-MoE in-memory02:04:09 - All of Google in one model02:12:43 - What’s missing from distillation02:18:03 - Open research, pros and cons02:24:54 - Going the distance This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Jeff Dean & Noam Shazeer — 25 years at Google: from PageRank to AGI

Third and final episode in the Paine trilogy!Chinese history is full of warlords constantly challenging the capital. How could Mao not only stay in power for decades, but not even face any insurgency?And how did Mao go from military genius to peacetime disaster - the patriotic hero who inflicted history’s worst human catastrophe on China? How can someone shrewd enough to win a civil war outnumbered 5 to 1 decide "let's have peasants make iron in their backyards" and "let's kill all the birds"?In her lecture and our Q&A, we cover the first nationwide famine in Chinese history; Mao's lasting influence on other insurgents; broken promises to minorities and peasantry; and what Taiwan means.Thanks so much to @Substack for running this in-person event!Note that Sarah is doing an AMA over the next couple days on Youtube; see the pinned comment.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform.SponsorToday’s episode is brought to you by Scale AI. Scale partners with the U.S. government to fuel America’s AI advantage through their data foundry. Scale recently introduced Defense Llama, Scale's latest solution available for military personnel. With Defense Llama, military personnel can harness the power of AI to plan military or intelligence operations and understand adversary vulnerabilities.If you’re interested in learning more on how Scale powers frontier AI capabilities, go to https://scale.com/dwarkesh. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Sarah Paine — How Mao conquered China (lecture & interview)

This is the second episode in the trilogy of a lectures by Professor Sarah Paine of the Naval War College.In this second episode, Prof Paine dissects the ideas and economics behind Japanese imperialism before and during WWII. We get into the oil shortage which caused the war; the unique culture of honor and death; the surprisingly chaotic chain of command. This is followed by a Q&A with me.Huge thanks to Substack for hosting this event!Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform.SponsorToday’s episode is brought to you by Scale AI. Scale partners with the U.S. government to fuel America’s AI advantage through their data foundry. Scale recently introduced Defense Llama, Scale's latest solution available for military personnel. With Defense Llama, military personnel can harness the power of AI to plan military or intelligence operations and understand adversary vulnerabilities.If you’re interested in learning more on how Scale powers frontier AI capabilities, go to scale.com/dwarkesh.Buy Sarah's Books!I highly, highly recommend both "The Wars for Asia, 1911–1949" and "The Japanese Empire: Grand Strategy from the Meiji Restoration to the Pacific War".Timestamps(0:00:00) - Lecture begins(0:06:58) - The code of the samurai(0:10:45) - Buddhism, Shinto, Confucianism(0:16:52) - Bushido as bad strategy(0:23:34) - Military theorists(0:33:42) - Strategic sins of omission(0:38:10) - Crippled logistics(0:40:58) - the Kwantung Army(0:43:31) - Inter-service communication(0:51:15) - Shattering Japanese morale(0:57:35) - Q&A begins(01:05:02) - Unusual brutality of WWII(01:11:30) - Embargo caused the war(01:16:48) - The liberation of China(01:22:02) - Could US have prevented war?(01:25:30) - Counterfactuals in history(01:27:46) - Japanese optimism(01:30:46) - Tech change and social change(01:38:22) - Hamming questions(01:44:31) - Do sanctions work?(01:50:07) - Backloaded mass death(01:54:09) - demilitarizing Japan(01:57:30) - Post-war alliances(02:03:46) - Inter-service rivalry This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Sarah Paine — Why Japan lost WWII (lecture & interview)

I’m thrilled to launch a new trilogy of double episodes: a lecture series by Professor Sarah Paine of the Naval War College, each followed by a deep Q&A.In this first episode, Prof Paine talks about key decisions by Khrushchev, Mao, Nehru, Bhutto, & Lyndon Johnson that shaped the whole dynamic of South Asia today. This is followed by a Q&A.Come for the spy bases, shoestring nukes, and insight about how great power politics impacts every region.Huge thanks to Substack for hosting this!Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform.SponsorsToday’s episode is brought to you by Scale AI. Scale partners with the U.S. government to fuel America’s AI advantage through their data foundry. The Air Force, Army, Defense Innovation Unit, and Chief Digital and Artificial Intelligence Office all trust Scale to equip their teams with AI-ready data and the technology to build powerful applications.Scale recently introduced Defense Llama, Scale's latest solution available for military personnel. With Defense Llama, military personnel can harness the power of AI to plan military or intelligence operations and understand adversary vulnerabilities.If you’re interested in learning more on how Scale powers frontier AI capabilities, go to scale.com/dwarkesh.Timestamps(00:00) - Intro(02:11) - Mao at war, 1949-51(05:40) - Pactomania and Sino-Soviet conflicts(14:42) - The Sino-Indian War(20:00) - Soviet peace in India-Pakistan(22:00) - US Aid and Alliances(26:14) - The difference with WWII(30:09) - The geopolitical map in 1904(35:10) - The US alienates Indira Gandhi(42:58) - Instruments of US power(53:41) - Carrier battle groups(1:02:41) - Q&A begins(1:04:31) - The appeal of the USSR(1:09:36) - The last communist premier(1:15:42) - India and China's lost opportunity(1:58:04) - Bismark's cunning(2:03:05) - Training US officers(2:07:03) - Cruelty in Russian history This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Sarah Paine — The war for India (lecture & interview)

I interviewed Tyler Cowen at the Progress Conference 2024. As always, I had a blast. This is my fourth interview with him – and yet I’m always hearing new stuff.We talked about why he thinks AI won't drive explosive economic growth, the real bottlenecks on world progress, him now writing for AIs instead of humans, and the difficult relationship between being cultured and fostering growth – among many other things in the full episode.Thanks to the Roots of Progress Institute (with special thanks to Jason Crawford and Heike Larson) for such a wonderful conference, and to FreeThink for the videography.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here.SponsorsI’m grateful to Tyler for volunteering to say a few words about Jane Street. It's the first time that a guest has participated in the sponsorship. I hope you can see why Tyler and I think so highly of Jane Street. To learn more about their open rules, go to janestreet.com/dwarkesh.Timestamps(00:00:00) Economic Growth and AI(00:14:57) Founder Mode and increasing variance(00:29:31) Effective Altruism and Progress Studies(00:33:05) What AI changes for Tyler(00:44:57) The slow diffusion of innovation(00:49:53) Stalin's library(00:52:19) DC vs SF vs EU This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Tyler Cowen — The #1 bottleneck to AI progress is humans

Adam Brown is a founder and lead of BlueShift with is cracking maths and reasoning at Google DeepMind and a theoretical physicist at Stanford.We discuss: destroying the light cone with vacuum decay, holographic principle, mining black holes, & what it would take to train LLMs that can make Einstein level conceptual breakthroughs.Stupefying, entertaining, & terrifying.Enjoy!Watch on YouTube, read the transcript, listen on Apple Podcasts, Spotify, or your favorite platform.Sponsors- Deepmind, Meta, Anthropic, and OpenAI, partner with Scale for high quality data to fuel post-training Publicly available data is running out - to keep developing smarter and smarter models, labs will need to rely on Scale’s data foundry, which combines subject matter experts with AI models to generate fresh data and break through the data wall. Learn more at scale.ai/dwarkesh.- Jane Street is looking to hire their next generation of leaders. Their deep learning team is looking for ML researchers, FPGA programmers, and CUDA programmers. Summer internships are open for just a few more weeks. If you want to stand out, take a crack at their new Kaggle competition. To learn more, go to janestreet.com/dwarkesh.- This episode is brought to you by Stripe, financial infrastructure for the internet. Millions of companies from Anthropic to Amazon use Stripe to accept payments, automate financial processes and grow their revenue.Timestamps(00:00:00) - Changing the laws of physics(00:26:05) - Why is our universe the way it is(00:37:30) - Making Einstein level AGI(01:00:31) - Physics stagnation and particle colliders(01:11:10) - Hitchhiking(01:29:00) - Nagasaki(01:36:19) - Adam’s career(01:43:25) - Mining black holes(01:59:42) - The holographic principle(02:23:25) - Philosophy of infinities(02:31:42) - Engineering constraints for future civilizations This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Adam Brown — Bubble universes, space elevators, & AdS/CFT

Gwern is a pseudonymous researcher and writer. He was one of the first people to see LLM scaling coming. If you've read his blog, you know he's one of the most interesting polymathic thinkers alive.In order to protect Gwern's anonymity, I proposed interviewing him in person, and having my friend Chris Painter voice over his words after. This amused him enough that he agreed.After the episode, I convinced Gwern to create a donation page where people can help sustain what he's up to. Please go here to contribute.Read the full transcript here.Sponsors:* Jane Street is looking to hire their next generation of leaders. Their deep learning team is looking for ML researchers, FPGA programmers, and CUDA programmers. Summer internships are open - if you want to stand out, take a crack at their new Kaggle competition. To learn more, go to janestreet.com/dwarkesh.* Turing provides complete post-training services for leading AI labs like OpenAI, Anthropic, Meta, and Gemini. They specialize in model evaluation, SFT, RLHF, and DPO to enhance models’ reasoning, coding, and multimodal capabilities. Learn more at turing.com/dwarkesh.* This episode is brought to you by Stripe, financial infrastructure for the internet. Millions of companies from Anthropic to Amazon use Stripe to accept payments, automate financial processes and grow their revenue.If you’re interested in advertising on the podcast, check out this page.Timestamps00:00:00 - Anonymity00:01:09 - Automating Steve Jobs00:04:38 - Isaac Newton's theory of progress00:06:36 - Grand theory of intelligence00:10:39 - Seeing scaling early00:21:04 - AGI Timelines00:22:54 - What to do in remaining 3 years until AGI00:26:29 - Influencing the shoggoth with writing00:30:50 - Human vs artificial intelligence00:33:52 - Rabbit holes00:38:48 - Hearing impairment00:43:00 - Wikipedia editing00:47:43 - Gwern.net00:50:20 - Counterfactual careers00:54:30 - Borges & literature01:01:32 - Gwern's intelligence and process01:11:03 - A day in the life of Gwern01:19:16 - Gwern's finances01:25:05 - The diversity of AI minds01:27:24 - GLP drugs and obesity01:31:08 - Drug experimentation01:33:40 - Parasocial relationships01:35:23 - Open rabbit holes This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Gwern — Anonymous writer who predicted AI trajectory on $12K/year salary

A bonanza on the semiconductor industry and hardware scaling to AGI by the end of the decade.Dylan Patel runs Semianalysis, the leading publication and research firm on AI hardware. Jon Y runs Asianometry, the world’s best YouTube channel on semiconductors and business history.* What Xi would do if he became scaling pilled* $ 1T+ in datacenter buildout by end of decadeWatch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Sponsors:* Jane Street is looking to hire their next generation of leaders. Their deep learning team is looking for FPGA programmers, CUDA programmers, and ML researchers. To learn more about their full time roles, internship, tech podcast, and upcoming Kaggle competition, go here.* This episode is brought to you by Stripe, financial infrastructure for the internet. Millions of companies from Anthropic to Amazon use Stripe to accept payments, automate financial processes and grow their revenue.If you’re interested in advertising on the podcast, check out this page.Timestamps00:00:00 – Xi's path to AGI00:04:20 – Liang Mong Song00:08:25 – How semiconductors get better00:11:16 – China can centralize compute00:18:50 – Export controls & sanctions00:32:51 – Huawei's intense culture00:38:51 – Why the semiconductor industry is so stratified00:40:58 – N2 should not exist00:45:53 – Taiwan invasion hypothetical00:49:21 – Mind-boggling complexity of semiconductors00:59:13 – Chip architecture design01:04:36 – Architectures lead to different AI models? China vs. US01:10:12 – Being head of compute at an AI lab01:16:24 – Scaling costs and power demand01:37:05 – Are we financing an AI bubble?01:50:20 – Starting Asianometry and SemiAnalysis02:06:10 – Opportunities in the semiconductor stack This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

@Asianometry & Dylan Patel — How the semiconductor industry actually works

Unless you understand the history of oil, you cannot understand the rise of America, WW1, WW2, secular stagnation, the Middle East, Ukraine, how Xi and Putin think, and basically anything else that's happened since 1860.It was a great honor to interview Daniel Yergin, the Pulitzer Prize winning author of The Prize - the best history of oil ever written (which makes it the best history of the 20th century ever written).Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Sponsors:This episode is brought to you by Stripe, financial infrastructure for the internet. Millions of companies from Anthropic to Amazon use Stripe to accept payments, automate financial processes and grow their revenue.This episode is brought to you by Suno, pioneers in AI-generated music. Suno's technology allows artists to experiment with melodic forms and structures in unprecedented ways. From chart-toppers to avant-garde compositions, Suno is redefining musical creativity. If you're an ML researcher passionate about shaping the future of music, email your resume to dwarkesh@suno.com.If you’re interested in advertising on the podcast, check out this page.Timestamps(00:00:00) – Beginning of the oil industry(00:13:37) – World War I & II(00:25:06) – The Middle East(00:47:04) – Yergin’s conversations with Putin & Modi(01:04:36) – Writing through stories(01:10:26) – The renewable energy transition This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Daniel Yergin — Oil destroyed Hitler, fracking destroyed Putin

I had no idea how wild human history was before chatting with the geneticist of ancient DNA David Reich.Human history has been again and again a story of one group figuring ‘something’ out, and then basically wiping everyone else out.From the tribe of 1k-10k modern humans who killed off all the other human species 70,000 years ago; to the Yamnaya horse nomads 5,000 years ago who killed off 90+% of (then) Europeans and also destroyed the Indus Valley.So much of what we thought we knew about human history is turning out to be wrong, from the ‘Out of Africa’ theory to the evolution of language, and this is all thanks to the research from David Reich’s lab.Buy David Reich’s fascinating book, Who We Are How We Got Here.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here.Follow me on Twitter for updates on future episodes.SponsorThis episode is brought to you by Stripe, financial infrastructure for the internet. Millions of companies from Anthropic to Amazon use Stripe to accept payments, automate financial processes and grow their revenue.If you’re interested in advertising on the podcast, check out this page.Timestamps(00:00:00) – Archaic and modern humans gene flow(00:20:24) – How early modern humans dominated the world(00:39:59) – How bubonic plague rewrote history(00:50:03) – Was agriculture terrible for humans?(00:59:28) – Yamnaya expansion and how populations collide(01:15:39) – “Lost civilizations” and our Neanderthal ancestry(01:31:32) – The DNA Challenge(01:41:38) – David’s career: the genetic vocation This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

David Reich — How one small tribe conquered the world 70,000 years ago

Chatted with Joe Carlsmith about whether we can trust power/techno-capital, how to not end up like Stalin in our urge to control the future, gentleness towards the artificial Other, and much more.Check out Joe's sequence on Otherness and Control in the Age of AGI here.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Sponsors:- Bland.ai is an AI agent that automates phone calls in any language, 24/7. Their technology uses "conversational pathways" for accurate, versatile communication across sales, operations, and customer support. You can try Bland yourself by calling 415-549-9654. Enterprises can get exclusive access to their advanced model at bland.ai/dwarkesh.- Stripe is financial infrastructure for the internet. Millions of companies from Anthropic to Amazon use Stripe to accept payments, automate financial processes and grow their revenue.If you’re interested in advertising on the podcast, check out this page.Timestamps:(00:00:00) - Understanding the Basic Alignment Story(00:44:04) - Monkeys Inventing Humans(00:46:43) - Nietzsche, C.S. Lewis, and AI(1:22:51) - How should we treat AIs(1:52:33) - Balancing Being a Humanist and a Scholar(2:05:02) - Explore exploit tradeoffs and AI This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Joe Carlsmith — Preventing an AI takeover

I talked with Patrick McKenzie (known online as patio11) about how a small team he ran over a Discord server got vaccines into Americans' arms: A story of broken incentives, outrageous incompetence, and how a few individuals with high agency saved 1000s of lives.Enjoy!Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here.Follow me on Twitter for updates on future episodes.SponsorThis episode is brought to you by Stripe, financial infrastructure for the internet. Millions of companies from Anthropic to Amazon use Stripe to accept payments, automate financial processes and grow their revenue.Timestamps(00:00:00) – Why hackers on Discord had to save thousands of lives(00:17:26) – How politics crippled vaccine distribution(00:38:19) – Fundraising for VaccinateCA(00:51:09) – Why tech needs to understand how government works(00:58:58) – What is crypto good for?(01:13:07) – How the US government leverages big tech to violate rights(01:24:36) – Can the US have nice things like Japan?(01:26:41) – Financial plumbing & money laundering: a how-not-to guide(01:37:42) – Maximizing your value: why some people negotiate better(01:42:14) – Are young people too busy playing Factorio to found startups?(01:57:30) – The need for a post-mortem This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Patrick McKenzie — Money laundering, big tech censorship, SBF & Japan

I chatted with Tony Blair about:- What he learned from Lee Kuan Yew- Intelligence agencies track record on Iraq & Ukraine- What he tells the dozens of world leaders who come seek advice from him- How much of a PM’s time is actually spent governing- What will AI’s July 1914 moment look like from inside the Cabinet?Enjoy!Watch the video on YouTube. Read the full transcript here.Follow me on Twitter for updates on future episodes.Sponsors- Prelude Security is the world’s leading cyber threat management automation platform. Prelude Detect quickly transforms threat intelligence into validated protections so organizations can know with certainty that their defenses will protect them against the latest threats. Prelude is backed by Sequoia Capital, Insight Partners, The MITRE Corporation, CrowdStrike, and other leading investors. Learn more here.- This episode is brought to you by Stripe, financial infrastructure for the internet. Millions of companies from Anthropic to Amazon use Stripe to accept payments, automate financial processes and grow their revenue.If you’re interested in advertising on the podcast, check out this page.Timestamps(00:00:00) – A prime minister’s constraints(00:04:12) – CEOs vs. politicians(00:10:31) – COVID, AI, & how government deals with crisis(00:21:24) – Learning from Lee Kuan Yew(00:27:37) – Foreign policy & intelligence(00:31:12) – How much leadership actually matters(00:35:34) – Private vs. public tech(00:39:14) – Advising global leaders(00:46:45) – The unipolar moment in the 90s This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Tony Blair — Why political leaders keep failing at major change

Here is my conversation with Francois Chollet and Mike Knoop on the $1 million ARC-AGI Prize they're launching today.I did a bunch of socratic grilling throughout, but Francois’s arguments about why LLMs won’t lead to AGI are very interesting and worth thinking through.It was really fun discussing/debating the cruxes. Enjoy!Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Timestamps(00:00:00) – The ARC benchmark(00:11:10) – Why LLMs struggle with ARC(00:19:00) – Skill vs intelligence(00:27:55) - Do we need “AGI” to automate most jobs?(00:48:28) – Future of AI progress: deep learning + program synthesis(01:00:40) – How Mike Knoop got nerd-sniped by ARC(01:08:37) – Million $ ARC Prize(01:10:33) – Resisting benchmark saturation(01:18:08) – ARC scores on frontier vs open source models(01:26:19) – Possible solutions to ARC Prize This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Francois Chollet — Why the biggest AI models can't solve simple puzzles

Chatted with my friend Leopold Aschenbrenner on the trillion dollar nationalized cluster, CCP espionage at AI labs, how unhobblings and scaling can lead to 2027 AGI, dangers of outsourcing clusters to Middle East, leaving OpenAI, and situational awareness.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here.Follow me on Twitter for updates on future episodes. Follow Leopold on Twitter.Timestamps(00:00:00) – The trillion-dollar cluster and unhobbling(00:20:31) – AI 2028: The return of history(00:40:26) – Espionage & American AI superiority(01:08:20) – Geopolitical implications of AI(01:31:23) – State-led vs. private-led AI(02:12:23) – Becoming Valedictorian of Columbia at 19(02:30:35) – What happened at OpenAI(02:45:11) – Accelerating AI research progress(03:25:58) – Alignment(03:41:26) – On Germany, and understanding foreign perspectives(03:57:04) – Dwarkesh’s immigration story and path to the podcast(04:07:58) – Launching an AGI hedge fund(04:19:14) – Lessons from WWII(04:29:08) – Coda: Frederick the Great This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history

Chatted with John Schulman (cofounded OpenAI and led ChatGPT creation) on how posttraining tames the shoggoth, and the nature of the progress to come...Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(00:00:00) - Pre-training, post-training, and future capabilities(00:16:55) - Plan for AGI 2025(00:29:18) - Teaching models to reason(00:39:45) - The Road to ChatGPT(00:51:07) - What makes for a good RL researcher?(00:59:53) - Keeping humans in the loop(01:14:11) - State of research, plateaus, and moatsSponsorsIf you’re interested in advertising on the podcast, fill out this form.* CommandBar is an AI user assistant that any software product can embed to non-annoyingly assist, support, and unleash their users. Used by forward-thinking CX, product, growth, and marketing teams. Learn more at commandbar.com. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

John Schulman (OpenAI Cofounder) — Reasoning, RLHF, & plan for 2027 AGI

Mark Zuckerberg on:- Llama 3- open sourcing towards AGI- custom silicon, synthetic data, & energy constraints on scaling- Caesar Augustus, intelligence explosion, bioweapons, $10b models, & much moreEnjoy!Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Human edited transcript with helpful links here.Timestamps(00:00:00) - Llama 3(00:08:32) - Coding on path to AGI(00:25:24) - Energy bottlenecks(00:33:20) - Is AI the most important technology ever?(00:37:21) - Dangers of open source(00:53:57) - Caesar Augustus and metaverse(01:04:53) - Open sourcing the $10b model & custom silicon(01:15:19) - Zuck as CEO of Google+SponsorsIf you’re interested in advertising on the podcast, fill out this form.* This episode is brought to you by Stripe, financial infrastructure for the internet. Millions of companies from Anthropic to Amazon use Stripe to accept payments, automate financial processes and grow their revenue. Learn more at stripe.com.* V7 Go is a tool to automate multimodal tasks using GenAI, reliably and at scale. Use code DWARKESH20 for 20% off on the pro plan. Learn more here.* CommandBar is an AI user assistant that any software product can embed to non-annoyingly assist, support, and unleash their users. Used by forward-thinking CX, product, growth, and marketing teams. Learn more at commandbar.com. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Mark Zuckerberg — Llama 3, $10B models, Caesar Augustus, & 1 GW datacenters

Had so much fun chatting with my good friends Trenton Bricken and Sholto Douglas on the podcast.No way to summarize it, except: This is the best context dump out there on how LLMs are trained, what capabilities they're likely to soon have, and what exactly is going on inside them.You would be shocked how much of what I know about this field, I've learned just from talking with them.To the extent that you've enjoyed my other AI interviews, now you know why.So excited to put this out. Enjoy! I certainly did :)Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. There's a transcript with links to all the papers the boys were throwing down - may help you follow along.Follow Trenton and Sholto on Twitter.Timestamps(00:00:00) - Long contexts(00:16:12) - Intelligence is just associations(00:32:35) - Intelligence explosion & great researchers(01:06:52) - Superposition & secret communication(01:22:34) - Agents & true reasoning(01:34:40) - How Sholto & Trenton got into AI research(02:07:16) - Are feature spaces the wrong way to think about intelligence?(02:21:12) - Will interp actually work on superhuman models(02:45:05) - Sholto’s technical challenge for the audience(03:03:57) - Rapid fire This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Sholto Douglas & Trenton Bricken — How LLMs actually think

Here is my episode with Demis Hassabis, CEO of Google DeepMindWe discuss:* Why scaling is an artform* Adding search, planning, & AlphaZero type training atop LLMs* Making sure rogue nations can't steal weights* The right way to align superhuman AIs and do an intelligence explosionWatch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Timestamps(0:00:00) - Nature of intelligence(0:05:56) - RL atop LLMs(0:16:31) - Scaling and alignment(0:24:13) - Timelines and intelligence explosion(0:28:42) - Gemini training(0:35:30) - Governance of superhuman AIs(0:40:42) - Safety, open source, and security of weights(0:47:00) - Multimodal and further progress(0:54:18) - Inside Google DeepMind This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Demis Hassabis — Scaling, superhuman AIs, AlphaZero atop LLMs, AlphaFold

We discuss:* what it takes to process $1 trillion/year* how to build multi-decade APIs, companies, and relationships* what's next for Stripe (increasing the GDP of the internet is quite an open ended prompt, and the Collison brothers are just getting started).Plus the amazing stuff they're doing at Arc Institute, the financial infrastructure for AI agents, playing devil's advocate against progress studies, and much more.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(00:00:00) - Advice for 20-30 year olds(00:12:12) - Progress studies(00:22:21) - Arc Institute(00:34:27) - AI & Fast Grants(00:43:46) - Stripe history(00:55:44) - Stripe Climate(01:01:39) - Beauty & APIs(01:11:51) - Financial innards(01:28:16) - Stripe culture & future(01:41:56) - Virtues of big businesses(01:51:41) - John This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Patrick Collison — Why Silicon Valley's most talented should leave

It was a great pleasure speaking with Tyler Cowen for the 3rd time.We discussed GOAT: Who is the Greatest Economist of all Time and Why Does it Matter?, especially in the context of how the insights of Hayek, Keynes, Smith, and other great economists help us make sense of AI, growth, animal spirits, prediction markets, alignment, central planning, and much more.The topics covered in this episode are too many to summarize. Hope you enjoy!Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(0:00:00) - John Maynard Keynes(00:17:16) - Controversy(00:25:02) - Fredrick von Hayek(00:47:41) - John Stuart Mill(00:52:41) - Adam Smith(00:58:31) - Coase, Schelling, & George(01:08:07) - Anarchy(01:13:16) - Cheap WMDs(01:23:18) - Technocracy & political philosophy(01:34:16) - AI & Scaling This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Tyler Cowen — Hayek, Keynes, & Smith on AI, animal spirits, anarchy, & growth

This is a narration of my blog post, Lessons from The Years of Lyndon Johnson by Robert Caro.You read the full post here: https://www.dwarkeshpatel.com/p/lyndon-johnsonListen on Apple Podcasts, Spotify, or any other podcast platform. Follow me on Twitter for updates on future posts and episodes. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Lessons from The Years of Lyndon Johnson by Robert Caro [Narration]

This is a narration of my blog post, Will scaling work?. You read the full post here: https://www.dwarkeshpatel.com/p/will-scaling-workListen on Apple Podcasts, Spotify, or any other podcast platform. Follow me on Twitter for updates on future posts and episodes.  This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Will scaling work? [Narration]

A true honor to speak with Jung Chang.She is the author of Wild Swans: Three Daughters of China (sold 15+ million copies worldwide) and Mao: The Unknown Story.We discuss:- what it was like growing up during the Cultural Revolution as the daughter of a denounced official- why the CCP continues to worship the biggest mass murderer in human history.- how exactly Communist totalitarianism was able to subjugate a billion people- why Chinese leaders like Xi and Deng who suffered from the Cultural Revolution don't condemn Mao- how Mao starved and killed 40 million people during The Great Leap Forward in order to exchange food for Soviet weaponsWild Swans is the most moving book I've ever read. It was a real privilege to speak with its author.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(00:00:00) - Growing up during Cultural Revolution(00:15:58) - Could officials have overthrown Mao?(00:34:09) - Great Leap Forward(00:48:12) - Modern support of Mao(01:03:24) - Life as peasant(01:21:30) - Psychology of communist society This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Jung Chang (Wild Swans author) — Living through history's largest man-made famine

Andrew Roberts is the world's best biographer and one of the leading historians of our time.We discussed* Churchill the applied historian,* Napoleon the startup founder,* why Nazi ideology cost Hitler WW2,* drones, reconnaissance, and other aspects of the future of war,* Iraq, Afghanistan, Korea, Ukraine, & Taiwan.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(00:00:00) - Post WW2 conflicts(00:10:57) - Ukraine(00:16:33) - How Truman Prevented Nuclear War(00:22:49) - Taiwan(00:27:15) - Churchill(00:35:11) - Gaza & future wars(00:39:05) - Could Hitler have won WW2?(00:48:00) - Surprise attacks(00:59:33) - Napoleon and startup founders(01:14:06) - Robert’s insane productivity This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Andrew Roberts — Why Hitler lost WWII, Churchill as applied historian, & Napoleon as startup founder

Here is my interview with Dominic Cummings on why Western governments are so dangerously broken, and how to fix them before an even more catastrophic crisis.Dominic was Chief Advisor to the Prime Minister during COVID, and before that, director of Vote Leave (which masterminded the 2016 Brexit referendum).Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(00:00:00) - One day in COVID…(00:08:26) - Why is government broken?(00:29:10) - Civil service(00:38:27) - Opportunity wasted?(00:49:35) - Rishi Sunak and Number 10 vs 11(00:55:13) - Cyber, nuclear, bio risks(01:02:04) - Intelligence & defense agencies(01:23:32) - Bismarck & Lee Kuan Yew(01:37:46) - How to fix the government?(01:56:43) - Taiwan(02:00:10) - Russia(02:07:12) - Bismarck’s career as an example of AI (mis)alignment(02:17:37) - Odyssean education This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Dominic Cummings - COVID, Brexit, & Fixing Western Governance

Paul Christiano is the world’s leading AI safety researcher. My full episode with him is out!We discuss:- Does he regret inventing RLHF, and is alignment necessarily dual-use?- Why he has relatively modest timelines (40% by 2040, 15% by 2030),- What do we want post-AGI world to look like (do we want to keep gods enslaved forever)?- Why he’s leading the push to get to labs develop responsible scaling policies, and what it would take to prevent an AI coup or bioweapon,- His current research into a new proof system, and how this could solve alignment by explaining model's behavior- and much more.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Open PhilanthropyOpen Philanthropy is currently hiring for twenty-two different roles to reduce catastrophic risks from fast-moving advances in AI and biotechnology, including grantmaking, research, and operations.For more information and to apply, please see the application: https://www.openphilanthropy.org/research/new-roles-on-our-gcr-team/The deadline to apply is November 9th; make sure to check out those roles before they close.Timestamps(00:00:00) - What do we want post-AGI world to look like?(00:24:25) - Timelines(00:45:28) - Evolution vs gradient descent(00:54:53) - Misalignment and takeover(01:17:23) - Is alignment dual-use?(01:31:38) - Responsible scaling policies(01:58:25) - Paul’s alignment research(02:35:01) - Will this revolutionize theoretical CS and math?(02:46:11) - How Paul invented RLHF(02:55:10) - Disagreements with Carl Shulman(03:01:53) - Long TSMC but not NVIDIA This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Paul Christiano — Preventing an AI takeover

I had a lot of fun chatting with Shane Legg - Founder and Chief AGI Scientist, Google DeepMind!We discuss:* Why he expects AGI around 2028* How to align superhuman models* What new architectures needed for AGI* Has Deepmind sped up capabilities or safety more?* Why multimodality will be next big landmark* and much moreWatch full episode on YouTube, Apple Podcasts, Spotify, or any other podcast platform. Read full transcript here.Timestamps(0:00:00) - Measuring AGI(0:11:41) - Do we need new architectures?(0:16:26) - Is search needed for creativity?(0:19:19) - Superhuman alignment(0:29:58) - Impact of Deepmind on safety vs capabilities(0:34:03) - Timelines(0:41:24) - Multimodality This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Shane Legg (DeepMind Founder) — 2028 AGI, superhuman alignment, new architectures

I had a lot of fun chatting with Grant Sanderson (who runs the excellent 3Blue1Brown YouTube channel) about:- Whether advanced math requires AGI- What careers should mathematically talented students pursue- Why Grant plans on doing a stint as a high school teacher- Tips for self teaching- Does Godel’s incompleteness theorem actually matter- Why are good explanations so hard to find?- And much moreWatch on YouTube. Listen on Spotify, Apple Podcasts, or any other podcast platform. Full transcript here.Timestamps(0:00:00) - Does winning math competitions require AGI?(0:08:24) - Where to allocate mathematical talent?(0:17:34) - Grant’s miracle year(0:26:44) - Prehistoric humans and math(0:33:33) - Why is a lot of math so new?(0:44:44) - Future of education(0:56:28) - Math helped me realize I wasn’t that smart(0:59:25) - Does Godel’s incompleteness theorem matter?(1:05:12) - How Grant makes videos(1:10:13) - Grant’s math exposition competition(1:20:44) - Self teaching This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Grant Sanderson (@3blue1brown) — Past, present, & future of mathematics

I learned so much from Sarah Paine, Professor of History and Strategy at the Naval War College.We discuss:- how continental vs maritime powers think and how this explains Xi & Putin's decisions- how a war with China over Taiwan would shake out and whether it could go nuclear- why the British Empire fell apart, why China went communist, how Hitler and Japan could have coordinated to win WW2, and whether Japanese occupation was good for Korea, Taiwan and Manchuria- plus other lessons from WW2, Cold War, and Sino-Japanese War- how to study history properly, and why leaders keep making the same mistakesIf you want to learn more, check out her books - they’re some of the best military history I’ve ever read.Watch on YouTube, listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript.Timestamps(0:00:00) - Grand strategy(0:11:59) - Death ground(0:23:19) - WW1(0:39:23) - Writing history(0:50:25) - Japan in WW2(0:59:58) - Ukraine(1:10:50) - Japan/Germany vs Iraq/Afghanistan occupation(1:21:25) - Chinese invasion of Taiwan(1:51:26) - Communists & Axis(2:08:34) - Continental vs maritime powers This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Sarah C. M. Paine — Why dictators keep making the same fatal mistake

Here is my conversation with Dario Amodei, CEO of Anthropic.Dario is hilarious and has fascinating takes on what these models are doing, why they scale so well, and what it will take to align them.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(00:00:00) - Introduction(00:01:00) - Scaling(00:15:46) - Language(00:22:58) - Economic Usefulness(00:38:05) - Bioterrorism(00:43:35) - Cybersecurity(00:47:19) - Alignment & mechanistic interpretability(00:57:43) - Does alignment research require scale?(01:05:30) - Misuse vs misalignment(01:09:06) - What if AI goes well?(01:11:05) - China(01:15:11) - How to think about alignment(01:31:31) - Is modern security good enough?(01:36:09) - Inefficiencies in training(01:45:53) - Anthropic’s Long Term Benefit Trust(01:51:18) - Is Claude conscious?(01:56:14) - Keeping a low profile This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Dario Amodei (Anthropic CEO) — The hidden pattern behind every AI breakthrough

A few weeks ago, I sat beside Andy Matuschak to record how he reads a textbook.Even though my own job is to learn things, I was shocked with how much more intense, painstaking, and effective his learning process was.So I asked if we could record a conversation about how he learns and a bunch of other topics:* How he identifies and interrogates his confusion (much harder than it seems, and requires an extremely effortful and slow pace)* Why memorization is essential to understanding and decision-making* How come some people (like Tyler Cowen) can integrate so much information without an explicit note taking or spaced repetition system.* How LLMs and video games will change education* How independent researchers and writers can make money* The balance of freedom and discipline in education* Why we produce fewer von Neumann-like prodigies nowadays* How multi-trillion dollar companies like Apple (where he was previously responsible for bedrock iOS features) manage to coordinate millions of different considerations (from the cost of different components to the needs of users, etc) into new products designed by 10s of 1000s of people.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.To see Andy’s process in action, check out the video where we record him studying a quantum physics textbook, talking aloud about his thought process, and using his memory system prototype to internalize the material.You can check out his website and personal notes, and follow him on Twitter.CometeerVisit cometeer.com/lunar for $20 off your first order on the best coffee of your life!If you want to sponsor an episode, contact me at dwarkesh.sanjay.patel@gmail.com.Timestamps(00:00:52) - Skillful reading(00:02:30) - Do people care about understanding?(00:06:52) - Structuring effective self-teaching(00:16:37) - Memory and forgetting(00:33:10) - Andy’s memory practice(00:40:07) - Intellectual stamina(00:44:27) - New media for learning (video, games, streaming)(00:58:51) - Schools are designed for the median student(01:05:12) - Is learning inherently miserable?(01:11:57) - How Andy would structure his kids’ education(01:30:00) - The usefulness of hypertext(01:41:22) - How computer tools enable iteration(01:50:44) - Monetizing public work(02:08:36) - Spaced repetition(02:10:16) - Andy’s personal website and notes(02:12:44) - Working at Apple(02:19:25) - Spaced repetition 2 This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Andy Matuschak — The reason most learning tools fail

The second half of my 7 hour conversation with Carl Shulman is out!My favorite part! And the one that had the biggest impact on my worldview.Here, Carl lays out how an AI takeover might happen:* AI can threaten mutually assured destruction from bioweapons,* use cyber attacks to take over physical infrastructure,* build mechanical armies,* spread seed AIs we can never exterminate,* offer tech and other advantages to collaborating countries, etcPlus we talk about a whole bunch of weird and interesting topics which Carl has thought about:* what is the far future best case scenario for humanity* what it would look like to have AI make thousands of years of intellectual progress in a month* how do we detect deception in superhuman models* does space warfare favor defense or offense* is a Malthusian state inevitable in the long run* why markets haven't priced in explosive economic growth* & much moreCarl also explains how he developed such a rigorous, thoughtful, and interdisciplinary model of the biggest problems in the world.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Catch part 1 hereTimestamps(0:00:00) - Intro (0:00:47) - AI takeover via cyber or bio (0:32:27) - Can we coordinate against AI? (0:53:49) - Human vs AI colonizers (1:04:55) - Probability of AI takeover (1:21:56) - Can we detect deception? (1:47:25) - Using AI to solve coordination problems (1:56:01) - Partial alignment (2:11:41) - AI far future (2:23:04) - Markets & other evidence (2:33:26) - Day in the life of Carl Shulman (2:47:05) - Space warfare, Malthusian long run, & other rapid fire This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Carl Shulman (Pt 2) — AI Takeover, bio & cyber attacks, detecting deception, & humanity's far future

In terms of the depth and range of topics, this episode is the best I’ve done.No part of my worldview is the same after talking with Carl Shulman. He's the most interesting intellectual you've never heard of.We ended up talking for 8 hours, so I'm splitting this episode into 2 parts.This part is about Carl’s model of an intelligence explosion, which integrates everything from:* how fast algorithmic progress & hardware improvements in AI are happening,* what primate evolution suggests about the scaling hypothesis,* how soon before AIs could do large parts of AI research themselves, and whether there would be faster and faster doublings of AI researchers,* how quickly robots produced from existing factories could take over the economy.We also discuss the odds of a takeover based on whether the AI is aligned before the intelligence explosion happens, and Carl explains why he’s more optimistic than Eliezer.The next part, which I’ll release next week, is about all the specific mechanisms of an AI takeover, plus a whole bunch of other galaxy brain stuff.Maybe 3 people in the world have thought as rigorously as Carl about so many interesting topics. This was a huge pleasure.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(00:00:00) - Intro(00:01:32) - Intelligence Explosion(00:18:03) - Can AIs do AI research?(00:39:00) - Primate evolution(01:03:30) - Forecasting AI progress(01:34:20) - After human-level AGI(02:08:39) - AI takeover scenarios This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Carl Shulman (Pt 1) — Intelligence explosion, primate evolution, robot doublings, & alignment

It was a tremendous honor & pleasure to interview Richard Rhodes, Pulitzer Prize winning author of The Making of the Atomic BombWe discuss- similarities between AI progress & Manhattan Project (developing a powerful, unprecedented, & potentially apocalyptic technology within an uncertain arms-race situation)- visiting starving former Soviet scientists during fall of Soviet Union- whether Oppenheimer was a spy, & consulting on the Nolan movie- living through WW2 as a child- odds of nuclear war in Ukraine, Taiwan, Pakistan, & North Korea- how the US pulled of such a massive secret wartime scientific & industrial projectWatch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(0:00:00) - Oppenheimer movie(0:06:22) - Was the bomb inevitable?(0:29:10) - Firebombing vs nuclear vs hydrogen bombs(0:49:44) - Stalin & the Soviet program(1:08:24) - Deterrence, disarmament, North Korea, Taiwan(1:33:12) - Oppenheimer as lab director(1:53:40) - AI progress vs Manhattan Project(1:59:50) - Living through WW2(2:16:45) - Secrecy(2:26:34) - Wisdom & war This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Richard Rhodes — The making of the atomic bomb

For 4 hours, I tried to come up reasons for why AI might not kill us all, and Eliezer Yudkowsky explained why I was wrong.We also discuss his call to halt AI, why LLMs make alignment harder, what it would take to save humanity, his millions of words of sci-fi, and much more.If you want to get to the crux of the conversation, fast forward to 2:35:00 through 3:43:54. Here we go through and debate the main reasons I still think doom is unlikely.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(0:00:00) - TIME article(0:09:06) - Are humans aligned?(0:37:35) - Large language models(1:07:15) - Can AIs help with alignment?(1:30:17) - Society’s response to AI(1:44:42) - Predictions (or lack thereof)(1:56:55) - Being Eliezer(2:13:06) - Othogonality(2:35:00) - Could alignment be easier than we think?(3:02:15) - What will AIs want?(3:43:54) - Writing fiction & whether rationality helps you win This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Eliezer Yudkowsky — Why AI will kill us, aligning LLMs, nature of intelligence, SciFi, & rationality

I went over to the OpenAI offices in San Fransisco to ask the Chief Scientist and cofounder of OpenAI, Ilya Sutskever, about:* time to AGI* leaks and spies* what's after generative models* post AGI futures* working with Microsoft and competing with Google* difficulty of aligning superhuman AIWatch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(00:00) - Time to AGI(05:57) - What’s after generative models?(10:57) - Data, models, and research(15:27) - Alignment(20:53) - Post AGI Future(26:56) - New ideas are overrated(36:22) - Is progress inevitable?(41:27) - Future Breakthroughs This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Ilya Sutskever (OpenAI Chief Scientist) — Why next-token prediction could surpass human intelligence

It is said that the two greatest problems of history are: how to account for the rise of Rome, and how to account for her fall. If so, then the volcanic ashes spewed by Mount Vesuvius in 79 AD - which entomb the cities of Pompeii and Herculaneum in South Italy - hold history’s greatest prize. For beneath those ashes lies the only salvageable library from the classical world.Nat Friedman was the CEO of Github form 2018 to 2021. Before that, he started and sold two companies - Ximian and Xamarin. He is also the founder of AI Grant and California YIMBY.And most recently, he has created and funded the Vesuvius Challenge - a million dollar prize for reading an unopened Herculaneum scroll for the very first time. If we can decipher these scrolls, we may be able to recover lost gospels, forgotten epics, and even missing works of Aristotle.We also discuss the future of open source and AI, running Github and building Copilot, and why EMH is a lie.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(0:00:00) - Vesuvius Challenge(0:30:00) - Finding points of leverage(0:37:39) - Open Source in AI(0:40:32) - Github Acquisition(0:50:18) - Copilot origin Story(1:11:47) - Nat.org(1:32:56) - Questions from Twitter This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Nat Friedman (Github CEO) — Reading ancient scrolls, open source, & AI

I flew out to Chicago to interview Brett Harrison, who is the former President of FTX US President and founder of Architect.In his first longform interview since the fall of FTX, he speak in great detail about his entire tenure there and about SBF’s dysfunctional leadership. He talks about how the inner circle of Gary Wang, Nishad Singh, and SBF mismanaged the company, controlled the codebase, got distracted by media, and even threatened him for his letter of resignation.In what was my favorite part of the interview, we also discuss his insights about the financial system from his decades of experience in the world's largest HFT firms.And we talk about Brett's new startup, Architect, as well as the general state of crypto post-FTX.Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(0:00:00) - Passive investing & HFT hacks(0:08:30) - Is Finance Zero-Sum?(0:18:38) - Interstellar Markets & Periodic Auctions(0:23:10) - Hiring & Programming at Jane Street(0:32:09) - Quant Culture(0:42:10) - FTX - Meeting Sam, Joining FTX US(0:58:20) - FTX - Accomplishments, Beginnings of Trouble(1:08:11) - FTX - SBF's Dysfunctional Leadership(1:26:53) - FTX - Alameda(1:33:50) - FTX - Leaving FTX, SBF"s Threats(1:45:45) - FTX - Collapse(1:53:10) - FTX - Lessons(2:04:34) - FTX - Regulators, & FTX Mafia(2:15:42) - Architect.xyz(2:30:10) - Institutional Interest & Uses of Crypto This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Brett Harrison — FTX US former president speaks out

My podcast with the brilliant Marc Andreessen is out!We discuss:* how AI will revolutionize software* whether NFTs are useless, & whether he should be funding flying cars instead* a16z's biggest vulnerabilities* the future of fusion, education, Twitter, venture, managerialism, & big techWatch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here. Follow me on Twitter for updates on future episodes.Timestamps(0:00:17) - Chewing glass(0:04:21) - AI(0:06:42) - Regrets(0:08:51) - Managerial capitalism(0:18:43) - 100 year fund(0:22:15) - Basic research(0:27:07) - $100b fund?(0:30:32) - Crypto debate(0:43:29) - Future of VC(0:50:20) - Founders(0:56:42) - a16z vulnerabilities(1:01:28) - Monetizing Twitter(1:07:09) - Future of big tech(1:14:07) - Is VC Overstaffed? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Marc Andreessen — AI, crypto, 1000 Elon Musks, regrets, vulnerabilities, & managerial revolution

Garett Jones is an economist at George Mason University and the author of The Cultural Transplant, Hive Mind, and 10% Less Democracy.This episode was fun and interesting throughout!He explains:* Why national IQ matters* How migrants bring their values to their new countries* Why we should have less democracy* How the Chinese are an unstoppable global force for free marketsWatch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform.Timestamps(00:00:00) - Intro(00:01:08) - Migrants Change Countries with Culture or Votes?(00:09:15) - Impact of Immigrants on Markets & Corruption(00:12:02) - 50% Open Borders?(00:16:54) - Chinese are Unstoppable Capitalists (00:21:39) - Innovation & Immigrants (00:24:53) - Open Borders for Migrants Equivalent to Americans?(00:28:54) - Let's Ignore Side Effects?(00:30:25) - Are Poor Countries Stuck?(00:32:26) - How Can Effective Altruists Increase National IQ(00:39:13) - Clone a million John von Neumann?(00:44:39) - Genetic Selection for IQ(00:47:02) - Democracy, Fed, FDA, & Presidential Power(00:49:42) - EU is a force for good?(00:55:12) - Why is America More Libertarian Than Median Voter?(00:56:19) - Is Ethnic Conflict a Short Run Problem?(00:59:38) - Bond Holder Democracy(01:04:57) - Mormonism(01:08:52) - Garett Jones's Immigration System(01:10:12) - Interviewing SBF This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Garett Jones — Immigration, national IQ, & less democracy

One of my best episodes ever. Lars Doucet is the author of Land is a Big Deal, a book about Georgism which has been praised by Vitalik Buterin, Scott Alexander, and Noah Smith. Sam Altman is the lead investor in his new startup, ValueBase.Talking with Lars completely changed how I think about who creates value in the world and who leeches off it.We go deep into the weeds on Georgism:* Why do even the wealthiest places in the world have poverty and homelessness, and why do rents increase as fast as wages?* Why are land-owners able to extract the profits that rightly belong to labor and capital?* How would taxing the value of land alleviate speculation, NIMBYism, and income and sales taxes?Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here.Follow Lars on Twitter. Follow me on Twitter.Timestamps(00:00:00) - Intro(00:01:11) - Georgism(00:03:16) - Metaverse Housing Crises(00:07:10) - Tax Leisure?(00:13:53) - Speculation & Frontiers(00:24:33) - Social Value of Search (00:33:13) - Will Georgism Destroy The Economy?(00:38:51) - The Economics of San Francisco(00:43:31) - Transfer from Landowners to Google?(00:46:47) - Asian Tigers and Land Reform(00:51:19) - Libertarian Georgism(00:55:42) - Crypto(00:57:16) - Transitioning to Georgism(01:02:56) - Lars's Startup & Land Assessment (01:15:12) - Big Tech(01:20:50) - Space(01:23:05) - Copyright(01:25:02) - Politics of Georgism(01:33:10) - Someone Is Always Collecting Rents This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com

Lars Doucet — Progress, poverty, Georgism, & why rent is too damn high

Holden Karnofsky is the co-CEO of Open Philanthropy and co-founder of GiveWell. He is also the author of one of the most interesting blogs on the internet, Cold Takes.We discuss:* Are we living in the most important century?* Does he regret OpenPhil’s 30 million dollar grant to OpenAI in 2016?* How does he think about AI, progress, digital people, & ethics?Highly recommend!Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform. Read the full transcript here.Timestamps(0:00:00) - Intro(0:00:58) - The Most Important Century(0:06:44) - The Weirdness of Our Time(0:21:20) - The Industrial Revolution (0:35:40) - AI Success Scenario(0:52:36) - Competition, Innovation , & AGI Bottlenecks(1:00:14) - Lock-in & Weak Points(1:06:04) - Predicting the Future(1:20:40) - Choosing Which Problem To Solve(1:26:56) - $30M OpenAI Investment(1:30:22) - Future Proof Ethics(1:37:28) - Integrity vs Utilitarianism(1:40:46) - Bayesian Mindset & Governance(1:46:56) - Career Advice This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com