{"slug": "no-error-signal-no-discovery", "title": "No Error Signal, No Discovery", "summary": "Tom Zahavy, a Google DeepMind researcher and co-author of AlphaProof, argues in an ICML 2026 position paper that large language models have mechanized induction and deduction but lack a mechanism for abduction, the mode of inference needed for scientific discovery. Using Einstein's development of general relativity as a case study, Zahavy contends that discovery requires inventing new hypotheses without an error signal, a capability current AI systems do not possess. The paper suggests that AI progress is verifier-gated, and abduction remains the bottleneck for true innovation.", "body_md": "[AI](https://sourcefeed.dev/c/ai)Article\n\n# No Error Signal, No Discovery\n\nAn AlphaProof co-author's ICML position paper pins down the one reasoning mode LLMs haven't mechanized.\n\n[Rachel Goldstein](https://sourcefeed.dev/u/rachel_goldstein)\n\nTom Zahavy is not an AI skeptic. He's a researcher at Google DeepMind and a co-author of [AlphaProof](https://deepmind.google/discover/blog/ai-solves-imo-problems-at-silver-medal-level/), the system that hit silver-medal standard on International Mathematical Olympiad problems in 2024. Which is exactly what makes his ICML 2026 position paper, [\"LLMs can't jump\"](https://openreview.net/forum?id=klU4737opt), worth your attention now that it's bouncing around the Hacker News front page. This isn't an outsider dunking on transformers. It's someone who helped mechanize formal reasoning drawing a precise line around what got mechanized — and what didn't.\n\nThe paper's claim, stripped of the philosophy-of-science scaffolding: machine learning has industrialized two of the three modes of inference and has no mechanism for the third. Induction — extracting rules from piles of examples — is what pretraining is. Deduction — deriving consequences from fixed premises — is falling fast, with AlphaProof's silver followed by gold-level IMO performance in 2025 and proof assistants like [Lean](https://lean-lang.org/) providing the verification substrate. The missing mode is abduction: inventing a new explanatory hypothesis when the data is thin and nothing is obviously broken.\n\n## A discovery with no error signal\n\nZahavy's case study is Einstein and general relativity, and the detail that matters is one most retellings skip: in 1907, Newtonian gravity wasn't failing. The equivalence of inertial and gravitational mass had been confirmed to one part in a billion. The lone anomaly — Mercury's drifting perihelion — was blamed on an undiscovered planet, \"Vulcan,\" not on the theory. In loss-function terms, Newton's residuals were near zero. There was no gradient pointing away from it.\n\nThat's the paper's sharpest move, because it kills the comfortable story that discovery is compression — find the simpler program that explains the data. A compression-driven optimizer sitting on 1907's data would patch Newton with a Vulcan parameter, not detonate absolute time and rebuild physics on curved spacetime, a step that makes the theory *more* complex before it gets simpler. Einstein's actual mechanism was a simulated sensation: an observer in free fall feels no gravity, so gravity and acceleration must be one phenomenon. Hypothesis first, evidence later — Eddington's eclipse measurements arrived years after the theory.\n\nZahavy is careful about what he's *not* claiming. He concedes that an LLM initialized with Einstein's 1915 postulates could plausibly grind out the field equations and the Mercury prediction — that's deduction, and we're good at automating it. The bottleneck is upstream, where the postulates come from.\n\n## The part worth stealing for your roadmap\n\nHere's the synthesis the paper gestures at but doesn't spell out: essentially every reliable LLM capability gain of the last two years has been verifier-gated. Coding agents work because tests and compilers are error signals. AlphaProof works because Lean rejects bad proofs. Reinforcement learning with verifiable rewards works precisely where a reward can be checked. Even DeepMind's AlphaEvolve, as the paper notes, optimizes against an explicit metric. Progress compounds wherever a gradient exists.\n\nAbduction is, by definition, the regime where no gradient exists — nothing is failing, so nothing tells the optimizer where to go. That gives you a brutally practical scoping question for any AI feature: **what's the verifier?** Migrating a codebase against a test suite, generating SQL you can execute, refactoring behind a type checker — verifier-shaped, automate aggressively and expect the ceiling to keep rising. Naming the right abstraction, deciding which product hypothesis to test, root-causing an incident that matches no known pattern — premise-shaped, and today's architectures give you no reason to expect autonomous competence. That split also explains why \"AI scientist\" products underwhelm: Zahavy's own read is that systems like Sakana's AI Scientist recombine existing symbolic concepts to optimize metrics — deduction and induction in a lab coat.\n\n## Where the argument overreaches\n\n\"Structurally incapable\" is a theorem-shaped claim supported by an essay. The paper offers no impossibility proof and no benchmark, and it generalizes from a single, maximally exceptional historical episode. The HN thread landed the obvious counterpunch: the \"LLMs can't X\" genre has a rough track record, usually resolving to \"they can, just badly at first.\" [ARC-AGI](https://arcprize.org/) already probes exactly this regime — inferring rules from a handful of examples, too sparse for statistics — and models have climbed it, expensively but measurably. Historians also note Lorentz and Poincaré had much of the mathematics before 1905; though that arguably strengthens Zahavy's point, since Einstein's leap was interpretive, not derivational.\n\nThe prescription is the weakest part. Zahavy's proposed fix — action-controllable world models in the lineage of DeepMind's Genie, giving agents a synthetic lab for counterfactual \"thought experiments\" — is a coherent research bet, but the empirical record for world models generating hypotheses (rather than pretty rollouts) is thin, and text-first labs keep shipping frontier reasoning without them. Zahavy himself moved to defuse the hype cycle, clarifying publicly that this is a personal position, not DeepMind's, and not a claim that LLMs can't contribute to discovery.\n\n## The taxonomy outlives the verdict\n\nMy call: the impossibility claim won't age well as stated — test-time search, tool loops, and whatever eats ARC-AGI next will keep blurring the boundary. But the Peirce-flavored taxonomy is immediately useful and most teams don't have one. It converts \"AI can't really think\" hand-waving into an operational question — is there an error signal here? — that cleanly predicts where agents already deliver and where they stall. Draw the line through your own backlog: everything downstream of a verifier is automation surface; everything upstream is where your senior people earn their comp, setting the axioms the machines then grind on. LLMs can't jump yet. They are, however, extraordinarily good at landing — once you tell them where.\n\n## Sources & further reading\n\n-\n[Position: LLMs Can't Jump](https://openreview.net/forum?id=klU4737opt)— openreview.net -\n[LLMs can't jump (full paper PDF)](https://www.tomzahavy.com/files/llms-cant-jump.pdf)— tomzahavy.com -\n[ICML 2026 Poster: Position: LLMs can't jump](https://icml.cc/virtual/2026/poster/67091)— icml.cc -\n[Position: LLMs Can't Jump - Hacker News discussion](https://news.ycombinator.com/item?id=49181083)— news.ycombinator.com -\n[Tom Zahavy's reflections on the LLMs Can't Jump paper](https://x.com/TZahavy/status/2082401499628376180)— x.com\n\n[Rachel Goldstein](https://sourcefeed.dev/u/rachel_goldstein)· Dev Tools Editor\n\nRachel has been embedded in the developer tooling ecosystem for nearly eight years, covering everything from IDE wars and package-manager drama to the quiet rise of AI-assisted coding. She has a soft spot for open-source maintainers and an unhealthy number of terminal emulators installed on a single laptop.\n\n## Discussion 0\n\nNo comments yet\n\nBe the first to weigh in.", "url": "https://wpnews.pro/news/no-error-signal-no-discovery", "canonical_source": "https://sourcefeed.dev/a/no-error-signal-no-discovery", "published_at": "2026-08-05 13:09:42+00:00", "updated_at": "2026-08-05 13:43:41.203343+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-research", "large-language-models"], "entities": ["Tom Zahavy", "Google DeepMind", "AlphaProof", "ICML", "Lean", "Einstein", "AlphaEvolve"], "alternates": {"html": "https://wpnews.pro/news/no-error-signal-no-discovery", "markdown": "https://wpnews.pro/news/no-error-signal-no-discovery.md", "text": "https://wpnews.pro/news/no-error-signal-no-discovery.txt", "jsonld": "https://wpnews.pro/news/no-error-signal-no-discovery.jsonld"}}