Subvolts take AI progress is faster than we think, with agents reaching expert breadth soon. Oct 23, 2025 - Developers
- Learn about the recipe for current AI progress and the frontier of rewards for agent training.
- Business
- Understand potential productivity gains from AI agents and how jobs might evolve.
- Everyday users
- Grasp the accelerating pace of AI development and its potential for future breakthroughs.
This interview with a key Anthropic researcher provides concrete timelines and insights into AI's rapid advancement.
Are We Misreading the AI Exponential? Julian Schrittwieser on Move 37 & Scaling RL (Anthropic)
Audio streams from Matt Turck. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
The MAD Podcast features Anthropic researcher Julian Schrittwieser discussing his views on AI's exponential progress. He predicts AI agents capable of full-day autonomous work by 2026 and expert-level breadth by 2027. The episode explores the combination of pre-training and reinforcement learning, the potential for AI to generate novel scientific breakthroughs, and the future impact on knowledge work.
Imagine AI getting smarter incredibly fast, like a student learning a year's worth of material every few months. This episode discusses how AI might soon do complex tasks all day on its own and even make major scientific discoveries, changing how we work and live.
This episode offers a deep dive into the future trajectory of AI from a leading Anthropic researcher. It's for anyone interested in understanding the pace of AI development beyond current hype, particularly those tracking advancements in agentic AI and potential scientific discovery. Schrittwieser's insights connect directly to the ongoing work at major labs like Google, OpenAI, and Meta, who are all pushing the boundaries of model capabilities and agentic systems.
- HostMatt Turck
- GuestJulian Schrittwiesertop AI researcher, Anthropic
- Guest: Julian Schrittwieser
- Title: Are We Misreading the AI Exponential? Julian Schrittwieser on Move 37 & Scaling RL (Anthropic)
- Show: The MAD Podcast
- Length: 70 minutes
- AI agents autonomous work prediction: by 2026
- AI expert-level breadth prediction: by 2027
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 com
- RL
- Julian Schrittwieser
- Reinforcement Learning, a way AI learns by trial and error to achieve goals.
- pre-training
- Julian Schrittwieser
- The initial learning phase for AI models on vast amounts of data.
- implicit world models
- Julian Schrittwieser
- Internal representations AI builds about how the world works.
- mechanistic interpretability
- Julian Schrittwieser
- Studying the inner workings of AI models to understand how they make decisions.
1 h 14
1 h 38
Questions people ask
- When might AI agents work autonomously for a full day?
- Julian Schrittwieser predicts AI agents could achieve full-day autonomous work by 2026.
- When could AI achieve expert-level breadth across tasks?
- Expert-level breadth in AI is anticipated by 2027, according to Schrittwieser.
- What combination of techniques drives current AI progress?
- The current leading approach combines pre-training with reinforcement learning (RL).
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Matt Turck.