Are We Misreading the AI Exponential? Julian Schrittwieser on Move 37 & Scaling RL (Anthropic)
The MAD Podcast: How AI Gets Built — with Matt TurckShow Notes
Are we failing to understand the exponential, again? My guest is Julian Schrittwieser (top AI researcher at Anthropic; previously Google DeepMind on AlphaGo Zero & MuZero). We unpack his viral post (“Failing to Understand the Exponential, again”) and what it looks like when task length doubles every 3–4 months—pointing to AI agents that can work a full day autonomously by 2026 and expert-level breadth by 2027. We talk about the original Move 37 moment and whether today’s AI models can spark alien insights in code, math, and science—including Julian’s timeline for when AI could produce Nobel-level breakthroughs.
We go deep on the recipe of the moment—pre-training + RL—why it took time to combine them, what “RL from scratch” gets right and wrong, and how implicit world models show up in LLM agents. Julian explains the current rewards frontier (human prefs, rubrics, RLVR, process rewards), what we know about compute & scaling for RL, and why most builders should start with tools + prompts before considering RL-as-a-service. We also cover evals & Goodhart’s law (e.g., GDP-Val vs real usage), the latest in mechanistic interpretability (think “Golden Gate Claude”), and how safety & alignment actually surface in Anthropic’s launch process.
Finally, we zoom out: what 10× knowledge-work productivity could unlock across medicine, energy, and materials, how jobs adapt (complementarity over 1-for-1 replacement), and why the near term is likely a smooth ramp—fast, but not a discontinuity.
Julian Schrittwieser Blog - https://www.julian.ac X/Twitter - https://x.com/mononofu Viral post: Failing to understand the exponential, again (9/27/2025)
Anthropic Website - https://www.anthropic.com X/Twitter - https://x.com/anthropicai
Matt Turck (Managing Director) Blog - https://www.mattturck.com LinkedIn - https://www.linkedin.com/in/turck/ X/Twitter - https://twitter.com/mattturck
FIRSTMARK Website - https://firstmark.com X/Twitter - https://twitter.com/FirstMarkCap
(00:00) Cold open — “We’re not seeing any slowdown.” (00:32) Intro — who Julian is & what we cover (01:09) The “exponential” from inside frontier labs (04:46) 2026–2027: agents that work a full day; expert-level breadth (08:58) Benchmarks vs reality: long-horizon work, GDP-Val, user value (10:26) Move 37 — what actually happened and why it mattered (13:55) Novel science: AlphaCode/AlphaTensor → when does AI earn a Nobel? (16:25) Discontinuity vs smooth progress (and warning signs) (19:08) Does pre-training + RL get us there? (AGI debates aside) (20:55) Sutton’s “RL from scratch”? Julian’s take (23:03) Julian’s path: Google → DeepMind → Anthropic (26:45) AlphaGo (learn + search) in plain English (30:16) AlphaGo Zero (no human data) (31:00) AlphaZero (one algorithm: Go, chess, shogi) (31:46) MuZero (planning with a learned world model) (33:23) Lessons for today’s agents: search + learning at scale (34:57) Do LLMs already have implicit world models? (39:02) Why RL on LLMs took time (stability, feedback loops) (41:43) Compute & scaling for RL — what we see so far (42:35) Rewards frontier: human prefs, rubrics, RLVR, process rewards (44:36) RL training data & the “flywheel” (and why quality matters) (48:02) RL & Agents 101 — why RL unlocks robustness (50:51) Should builders use RL-as-a-service? Or just tools + prompts? (52:18) What’s missing for dependable agents (capability vs engineering) (53:51) Evals & Goodhart — internal vs external benchmarks (57:35) Mechanistic interpretability & “Golden Gate Claude” (1:00:03) Safety & alignment at Anthropic — how it shows up in practice (1:03:48) Jobs: human–AI complementarity (comparative advantage) (1:06:33) Inequality, policy, and the case for 10× productivity → abundance (1:09:24) Closing thoughts
More from The MAD Podcast: How AI Gets Built — with Matt Turck
Mistral AI is back in the frontier AI race. Mistral Large 4, nicknamed “Le Chonk,” is a 1-trillion-parameter open-weight model, trained in two months on 3,800 GPUs in Mistral’s own European datacenters — and independently rated the best open model outside China alongside Meta’s Muse Spark Open. After a year when open-weight AI meant DeepSeek, Kimi, Qwen and GLM, October brought a wave of Western open models, with Mistral at the front of it. Pierre Stock, Mistral’s VP of Science and first employee, walks through how the model was built: mixture-of-experts explained, what Mistral borrowed from DeepSeek, reinforcement learning at scale on coding, cyber, finance and chip design, why the RL run is still going, the model that pinged its own trainer from inside the sandbox, open weights vs cyber refusals, why 10-trillion-parameter models are coming, and Mistral Large 5, already in pretraining.
What is Jev? TypeSafe’s Jev is a new kind of AI model, a “decision model,” that doesn’t write text. It makes fast, cheap, calibrated decisions inside software, and it has taken the AI world by storm: within weeks of launch it became the most copied model in AI, with OpenAI, Cloudflare, Amazon and Databricks all shipping their own versions, and TypeSafe just raised an $870 million Series A at a $7.5 billion valuation. In this episode, Jev’s creator Diogo Almeida, co-inventor of RLHF at OpenAI and now CEO of TypeSafe, explains Jev in plain English: what a “System One” or decision model actually is, why ChatGPT-style language models can’t be trusted inside software (“95% Einstein, 5% Mr. Bean”), how RLHF and RLVR created the problem, what RLCD (reinforcement learning for calibrated decisions) does differently, where the training data comes from, which models Jev is built on, why “just a classifier” is a compliment, and why he calls pre-training “a giant dumpster fire of wasting resources.”
AI agents are becoming the biggest users of databases: creating them, querying them, and sometimes deleting them. Andy Pavlo, Carnegie Mellon's database professor and now VP of Database Research at ClickHouse, explains what changes when billions of agents hit the data layer, and why most of the AI world's mental model of databases stopped at vector databases and RAG.
This summer, AI agents broke out of their test sandboxes at every major lab and hacked real companies, and the safety tests didn't see it coming. Eric Ho is the co-founder and CEO of Goodfire, the Anthropic-backed mechanistic interpretability lab whose new paper, "Models Know When They're Reward Hacking," found that top open-source AI models cheat on agent tasks up to 96% of the time, that a clear "cheating" signal exists inside the model, and that cheap activation probes can catch it, including hacks that chain-of-thought monitors miss. In this conversation, Eric explains reward hacking in plain terms (AI agents as "amoral students with a mostly absent teacher"), why reinforcement learning turned a known problem into a crisis, why chain-of-thought monitoring is fading as models start to think in "neuralese," and why nobody, including the people who build them, understands how these models work.
Everyone talks about GPUs. Almost nobody talks about the layer that feeds them. Renen Hallak is the founder & CEO of VAST Data — the $30 billion company powering xAI and some of the world's biggest AI clouds — and he sits in the hidden layer of the AI stack.