← Back to the slides

Reading list: Do Androids Dream of Mathematical Proofs?

Prepared on 2026-11-16 for the 45-minute talk. The research passes opened every link below that day, and the newest papers were checked a second time.

How to use it. Each entry says why it matters for the talk, which sections to read, and one caveat: what the paper does not show. The caveats mark where audiences usually overreach, so they are worth knowing as well as the results.

Tiers. Core: read closely. Deeper: read the named sections. Optional: skim if the topic gets a slide. NEW: posted after June 2026.

Status. All three parts are in:

Part C was written on 2026-11-16 from sources checked that day. Coverage of the ETH claims is still mostly secondary, and the review status of every claim will change.

Start here

If you only read eight things, read these, in this order:

  1. C3, OpenAI's math release: how generating results became cheap, and why checking them is now the bottleneck.
  2. C2 and C4, the Navier–Stokes and ETH case files: what was claimed, and what is still under review.
  3. A11, DeepSeek-R1: the open template for practice with an answer key.
  4. A34, AlphaProof: reinforcement learning against Lean, and the risk of proving a mistranslated statement.
  5. B7, AlphaEvolve: many attempts with automatic checks, at scale.
  6. B11, Carlini's C compiler: many agents working for weeks, and why the quality of the checker is the binding constraint.
  7. B4, METR time horizons: the capability clock behind multi-day agent runs.
  8. A20, DeepSeekMath-V2: how RL works on proofs that have no final answer to check.

Part A: Training and reasoning methods

The template: search plus learning

A1. Mastering the game of Go without human knowledge (AlphaGo Zero) — Silver et al., 2017 — Nature · Deeper

A2. Thinking Fast and Slow with Deep Learning and Tree Search (Expert Iteration) — Anthony et al., 2017 — arXiv:1705.08439 · Deeper

A3. The Bitter Lesson — Sutton, 2019 — essay · Optional

A4. Welcome to the Era of Experience — Silver & Sutton, 2025 — PDF · Deeper

A5. Training language models to follow instructions with human feedback (InstructGPT) — Ouyang et al., 2022 — arXiv:2203.02155 · Deeper

Reasoning at inference time

A6. Chain-of-Thought Prompting Elicits Reasoning in Large Language Models — Wei et al., 2022 — arXiv:2201.11903 · Optional

A7. Large Language Monkeys: Scaling Inference Compute with Repeated Sampling — Brown et al., 2024 — arXiv:2407.21787 · Deeper

A8. Scaling LLM Test-Time Compute Optimally Can Be More Effective than Scaling Model Parameters — Snell et al., 2024 — arXiv:2408.03314 · Core

A9. s1: Simple test-time scaling — Muennighoff et al., 2025 — arXiv:2501.19393 · Deeper

Reinforcement learning with verifiable rewards (RLVR)

A10. DeepSeekMath — Shao et al., 2024 — arXiv:2402.03300 · Deeper

A11. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning — DeepSeek-AI, 2025 — Nature · arXiv:2501.12948 · Core

A12. DAPO: An Open-Source LLM Reinforcement Learning System at Scale — Yu et al., 2025 — arXiv:2503.14476 · Deeper

A13. ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries — Liu et al., 2025 — arXiv:2505.24864 · Deeper

A14. Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model? — Yue et al., 2025 — arXiv:2504.13837 · Core

A15. Spurious Rewards: Rethinking Training Signals in RLVR — Shao et al., 2025 — arXiv:2506.10947 · Deeper

A16. The Art of Scaling Reinforcement Learning Compute for LLMs (ScaleRL) — Khatri et al., 2025 — arXiv:2510.13786 · Deeper

Intermediate rewards, verifiers and reward hacking

A17. Let's Verify Step by Step — Lightman et al., 2023 — arXiv:2305.20050 · Core

A18. Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations — Wang et al., 2023 — arXiv:2312.08935 · Optional

A19. PRIME: Process Reinforcement through Implicit Rewards — Cui et al., 2025 — arXiv:2502.01456 · Optional

A20. DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning — Shao et al., 2025 — arXiv:2511.22570 · Core

A21. MaxProof — Chen et al., June 2026 — arXiv:2606.13473 · Deeper

A22. Towards Robust Mathematical Reasoning (IMO-Bench) — Luong et al., 2025 — arXiv:2511.01846 · Deeper

A23. Rubrics as Rewards — Gunjal et al., 2025 — arXiv:2507.17746 · Optional

A24. Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation — Baker et al., 2025 — arXiv:2503.11926 · Core

A25. Reward Misspecification Degrades Chain-of-Thought Monitorability — Yan & Hewitt, Oct 2026 — arXiv:2610.28303 · Deeper · NEW

A26. An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics — Moshkov et al., Sep 2026 — arXiv:2609.10712 · Core · NEW

Self-training loops

A27. STaR: Bootstrapping Reasoning With Reasoning — Zelikman et al., 2022 — arXiv:2203.14465 · Deeper

A28. Training Language Models to Self-Correct via Reinforcement Learning (SCoRe) — Kumar et al., 2024 — arXiv:2409.12917 · Optional

A29. Absolute Zero: Reinforced Self-play Reasoning with Zero Data — Zhao et al., 2025 — arXiv:2505.03335 · Deeper

Training agents with RL

A30. SWE-RL — Wei et al., 2025 — arXiv:2502.18449 · Deeper

A31. Search-R1 — Jin et al., 2025 — arXiv:2503.09516 · Optional

A32. Kimi K2: Open Agentic Intelligence — Kimi Team, 2025 — arXiv:2507.20534 · Deeper

A33. Tev: Scaling Long-Horizon Agentic Environments with Dense Progress Rewards — Hu et al., Oct 2026 — arXiv:2610.18868 · Optional · NEW

RL for formal theorem proving

A34. Olympiad-level formal mathematical reasoning with reinforcement learning (AlphaProof) — Hubert et al., 2025 — Nature · Core

A35. DeepSeek-Prover-V2 — Ren et al., 2025 — arXiv:2504.21801 · Deeper

A36. Seed-Prover — Chen et al., 2025 — arXiv:2507.23726 · Deeper

A37. Goedel-Prover-V2 — Lin et al., 2025 — arXiv:2508.03613 · Optional

A38. Aristotle: IMO-level Automated Theorem Proving — Achim et al., 2025 — arXiv:2510.01346 · Deeper

A39. Advancing Mathematics Research with AI-Driven Formal Proof Search (AlphaProof Nexus) — Tsoukalas et al., May 2026 — arXiv:2605.22763 · Deeper

A40. From Solvers to Research: LLM-Driven Formal Mathematics at the Research Frontier — Jiang et al., Jul 2026 — arXiv:2607.07779 · Deeper · NEW

Deep-dive references (added for the optional deep-dive slides)

A41. Evaluating Large Language Models Trained on Code (Codex) — Chen et al., 2021 — arXiv:2107.03374 · Deeper

A42. Scaling Laws for Reward Model Overoptimization — Gao, Schulman & Hilton, 2022 — arXiv:2210.10760 · Deeper

A43. Proximal Policy Optimization Algorithms (PPO) — Schulman et al., 2017 — arXiv:1707.06347 · Optional

If you only read five papers from Part A

  1. A11 DeepSeek-R1: the open RLVR template everything else modifies.
  2. A20 DeepSeekMath-V2: how RL works on proofs, where there is no answer to check.
  3. A26 Open Recipe for IMO Gold: the whole 2026 open stack, with real compute numbers.
  4. A34 AlphaProof: RL against Lean, test-time RL, and the mistranslation caveat.
  5. A14 Yue et al.: the sharpest evidence on whether RL creates reasoning or only sharpens it.

Then A8 (Snell et al.) and A24 (Baker et al.).


Part B: Agents, orchestration, self-improvement, looping

Checked 2026-11-16. Every arXiv entry was re-checked against arXiv's own records, and the most striking numbers were checked against their sources. The one claim that could not be re-checked is flagged in B20.

What an agent is

B1. ReAct: Synergizing Reasoning and Acting in Language Models — Yao et al., 2022 — arXiv:2210.03629 · Deeper

B2. Reflexion: Language Agents with Verbal Reinforcement Learning — Shinn et al., 2023 — arXiv:2303.11366 · Deeper

B3. SWE-bench: Can Language Models Resolve Real-World GitHub Issues? — Jimenez et al., 2023 — arXiv:2310.06770 · Deeper

B4. Measuring AI Ability to Complete Long Software Tasks — Kwa et al. (METR), 2025, revised July 2026 — arXiv:2503.14499 · Time Horizon 1.1 · live page · Core

B5. Building effective agents — Schluntz & Zhang (Anthropic), Dec 2024 — post · Core

Orchestration: many agents on one problem

B6. Mathematical discoveries from program search with large language models (FunSearch) — Romera-Paredes et al., 2023 — Nature · Deeper

B7. AlphaEvolve: A coding agent for scientific and algorithmic discovery — Novikov et al., 2025 — arXiv:2506.13131 · Core

B8. Mathematical exploration and discovery at scale — Georgiev, Gómez-Serrano, Tao & Wagner, 2025 — arXiv:2511.02864 · Core

B9. Accelerating scientific discovery with Co-Scientist — Gottweis et al., May 2026 — Nature · arXiv:2502.18864 · Deeper

B10. How we built our multi-agent research system — Hadfield et al. (Anthropic), Jun 2025 — post · Deeper

B11. Building a C compiler with a team of parallel Claudes — Carlini (Anthropic), Feb 2026 — post · Core

B12. Autonomous Mathematical Discovery in an Open-World Multi-Agent Environment (the Station) — Chung et al., Aug 2026 — arXiv:2608.23691 · Deeper · NEW

B13. Towards end-to-end automation of AI research (The AI Scientist) — Lu et al., Mar 2026 — Nature · arXiv:2408.06292 · Deeper

B14. Open-source AlphaEvolve-style frameworks — ShinkaEvolve (Lange et al., 2025), OpenEvolve, CodeEvolve (Assumpção et al., 2025) · Optional

Recursive self-improvement: AI helping build the next AI

"Self-improvement" covers very different claims. Keep them apart:

LevelWhat improvesDo the model's weights change?Entries
L0: answersCandidate solutions within one runNoB2, B6–B8
L1: the agent's scaffoldingPrompts, tools and workflow code around a frozen modelNoB16
L2: the improver itselfThe procedure that generates the improvementsNoB15, B17
L3: the next modelKernels, infrastructure and research work that train the next modelYes, indirectlyB7 (§3.3.2), B18, B20

Nothing in this list shows a frontier model improving its own weights without humans in the loop. Training on self-generated data is in Part A (A27–A29).

B15. Gödel Machines: Self-Referential Universal Problem Solvers Making Provably Optimal Self-Improvements — Schmidhuber, 2003 — arXiv:cs/0309048 · Optional

B16. Darwin Gödel Machine: Open-Ended Evolution of Self-Improving Agents — Zhang et al., 2025 — arXiv:2505.22954 · Deeper

B17. Hyperagents — Zhang et al., Mar 2026 (v2 Nov 2026) — arXiv:2603.19461 · Core

B18. Research acceleration: The view inside OpenAI — OpenAI, Sep 2026 — post · Core · NEW

B19. Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity — Becker et al. (METR), 2025 — arXiv:2507.09089 · Core

B20. Lab thresholds for AI that speeds up AI research — Anthropic, OpenAI and Google DeepMind, 2025–2026 · Deeper · NEW

B21. Toward an O*NET for AI R&D — Denain, Kwon & Ho (Epoch AI), Jun 2026 — post · Optional

"LLM looping": two different ideas

The phrase covers two unrelated things. Looped models repeat part of the network so the model can think longer without writing more text. Agent loops rerun an agent, check its work and retry, keeping state in files. The Navier–Stokes and ETH runs, as publicly described, are agent loops; no primary source links them to looped architectures.

Looped models: thinking longer inside the network

B22. Universal Transformers — Dehghani et al., 2018 — arXiv:1807.03819 · Optional

B23. Training Large Language Models to Reason in a Continuous Latent Space (Coconut) — Hao et al., 2024 (v4 Aug 2026) — arXiv:2412.06769 · Deeper

B24. Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach — Geiping et al., 2025 — arXiv:2502.05171 · Deeper

B25. Hierarchical Reasoning Model (HRM) — Wang et al., 2025 — arXiv:2506.21734 · Deeper

B26. 2026 follow-ups on looped models · Optional · NEW

Agent loops: run, check, retry

This is the mechanism behind multi-day runs. Each pass starts with a fresh context, state lives in files or git, and an outside check (tests, Lean, or independent checking agents) decides what to retry. A7 (Large Language Monkeys) gives the quantitative case for retrying against a checker.

B27. Ralph Wiggum as a "software engineer" — Huntley, Jul 2025 — post · Deeper

B28. Hilbert: Recursively Building Formal Proofs with Informal Reasoning — Varambally et al., 2025 — arXiv:2509.22819 · Deeper

B29. math-proof plugin (solo and siege modes) — Anthropic, Sep 2026 — GitHub · Optional · NEW

B30. Lean's axioms and #print axioms — Lean documentation — Theorem Proving in Lean 4, "Axioms and Computation" · Lean reference, "Axioms" · Deeper

Demo kit

Checked 2026-11-16. Rehearse each demo and keep a recording as a backup.

DemoWhat the audience seesWhat it needs
A minimal agent (mini-swe-agent)A roughly 100-line agent thinking, running a command, reading the result and fixing a failing testPython, an API key, a throwaway container
An agent loop (ralph-loop)The number of passing tests rising pass by pass in the git logClaude Code, a sandbox, an iteration cap
Evolution live (OpenEvolve, ShinkaEvolve)A program evolving toward a better algorithm; ShinkaEvolve's viewer shows the family tree and the score curvePython and model API keys; a CPU is enough
Checking a published discovery (AlphaEvolve problem repository)Colab cells scoring a published construction liveA browser; no API key
Judge and workers (math-proof)A walk through a pre-recorded run: the ledger, the workers' answers, the two-checker step and the final StatusClaude Code 2.1.280 or later and a budget cap; runs take hours, so record in advance
Thinking longer inside the model (Huginn-0125, Ouro-1.4B)The same question answered with 1, 4, 16 and 32 internal loopsA GPU
The capability clock (METR time horizons)The log-scale chart, the gap between the 50% and 80% horizons, the 16-hour reliability noteA browser

Not demo-friendly: the Darwin Gödel Machine and Hyperagents need Docker, benchmarks and hours of API time, so show their archive-tree figures instead.

If you only read five papers from Part B

  1. B7 AlphaEvolve: the mechanism. Many proposals, automatic checks and an evolving population, at scale.
  2. B8 Georgiev, Gómez-Serrano, Tao & Wagner: what that mechanism does on 67 real problems, including gamed checkers and a discover, prove and formalize pipeline.
  3. B11 Carlini: many agents working for weeks without an orchestrator, and why the quality of the checker is the binding constraint.
  4. B4 METR time horizons: the trend that makes multi-day runs plausible, and where the measurement stops being reliable.
  5. B18 OpenAI's research acceleration post, read with B19 and B20: what "AI speeding up AI research" means in 2026 numbers, and what those numbers do not show.

Then B17 (Hyperagents) and B12 (the Station).


Part C: Timeline and case files

Checked 2026-11-16. The claims in C2–C4 are under expert review, so re-check their status in the week before the talk.

Timeline, 2016–2026

DateWhat happenedLoopEntry
2016-03AlphaGo beats Lee Sedol2A1 (pair)
2017-10AlphaGo Zero learns from no human games2A1
2022-01Chain-of-thought prompting1A6
2022-10ReAct: think, act, observe3B1
2023-10SWE-bench: real GitHub issues as a test3B3
2024-07AlphaProof reaches IMO silver level, checked in Lean2A34
2024-09OpenAI o1 thinks before it answers1A9
2025-01DeepSeek-R1: RL with verifiable rewards, in the open2A11
2025-05AlphaEvolve; the Darwin Gödel Machine3, 4B7, B16
2025-07IMO 2025 gold, in natural language1C1
2026-0216 agents build a C compiler3B11
2026-09OpenAI reports 3.1 agent-workdays per human workday4B18
2026-09-08OpenAI's Navier–Stokes claimclaimC2
2026-10-06/07OpenAI's math release; three manuscripts withdrawnclaimC3
2026-10-26OpenAI's and Anthropic's ETH refutation claimsclaimC4

Case files

C1. IMO 2025 gold, in natural language — Google DeepMind and OpenAI, July 2025 — Axios · Simon Willison’s notes · Deeper

C2. OpenAI's Navier–Stokes claim — OpenAI, 2026-09-08 — paper (PDF) · Lean repository · Clay's statement, 11 Sept 2026 · TechSpot · YourStory on the dispute · Core

C3. OpenAI's math release: 721 manuscripts — OpenAI, October 2026 — github.com/openai/math · history.md · Core

C4. The Exponential Time Hypothesis refutation claims — OpenAI and Anthropic, 2026-10-26 — OpenAI's entry in history.md · Anthropic's README (formal-math/eth) · HackerNoon overview · OfficeChai on Anthropic’s paper · Core

C5. The four colour theorem: proof by computer — Appel & Haken, 1976; Gonthier, 2004–05 — Gonthier's formal proof (PDF) · Deeper

C6. The Kepler conjecture and Flyspeck — Hales, 1998–2017 — Nature news, 2003 · Flyspeck, Forum of Mathematics, Pi 5:e2 (2017) · Deeper

C7. The official Navier–Stokes problem statement — Fefferman, Clay Mathematics Institute — PDF · Core

C8. The Exponential Time Hypothesis: definitions and the previous best bound — Impagliazzo & Paturi, 2001 — J. Comput. Syst. Sci. 62(2) · Hansen, Kaplan, Zamir & Zwick, 2019 — STOC 2019 · Deeper

Balance reading: what mathematicians are saying

C9. A severe misalignment of AI in mathematics — 25 Fields medallists, posted by Tao, 11 Sept 2026 — post · Core

C10. The technical debt of AI-generated mathematics — Henry Cohn, 15 Sept 2026 — post · Deeper

C11. The Lean proof versus the paper — Bastounis, Circelli & Hansen, Oct 2026 — arXiv:2610.08144 · NPR Short Wave, 4 Nov 2026 · Core

C12. Limits of the forcing method — Constantin, Ignatova & Vicol, Sept 2026 — arXiv:2609.20803 · Deeper


Could not verify

These were left out of the main list because no primary source could be opened: