I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.
Schema: Frontier Models with the Right Harness Achieve ~99% on ARC-AGI-3 PublicSchema's claim that frontier models with the right harness achieve ~99% on ARC-AGI-3 is directly challenged by the thesis that LLMs lack manipulative abduction and physical simulation — benchmark success may not reflect genuine inventive or scientific reasoning capability.
Ten Advances In MathematicsTen Advances in Mathematics celebrates AI progress in mathematical domains, but the manipulative abduction thesis argues that true scientific invention — including mathematical discovery — requires cognitive processes LLMs structurally lack, not just more capability.
Supports
Brad Porter: We Saw This With Robotics at Amazon TooBrad Porter's robotics-at-Amazon perspective implicitly involves the gap between language-level reasoning and physical manipulation — the 'LLMs can't jump' thesis about needing physical simulation maps onto robotics practitioners' observations about grounding AI in physical action.
Intelligence is not the main bottleneckBoth argue that current AI/LLM capabilities miss something fundamental — the 'LLMs can't jump' thesis that physical simulation and manipulative abduction are required for invention aligns with the claim that intelligence itself isn't the key bottleneck, suggesting different cognitive primitives are missing.
Leiden Declaration on Artificial Intelligence and MathematicsThe Leiden Declaration on AI and Mathematics likely raises concerns about LLM limitations in genuine mathematical discovery; the 'manipulative abduction' thesis directly explains WHY LLMs struggle with novel scientific/mathematical invention rather than pattern-matched reasoning.
Related to
Daphne Koller: Insitro, AI-Driven Drug Discovery, and the Virtual Human PlatformDaphne Koller's AI-driven drug discovery via 'Virtual Human Platform' is a domain where manipulative abduction and physical simulation are explicitly required — the item's thesis has direct implications for the limits of LLM contribution to scientific drug discovery.
Supported by
Humanising LLM Outputs is DumbBoth argue against anthropomorphizing or over-attributing human-like qualities to LLMs — 'LLMs can't jump' challenges capability assumptions while this piece challenges the framing of making outputs 'sound human'
Understanding Embeddings in Language Models'LLMs can't jump' likely explores limitations of language models; understanding how embeddings encode semantic space helps explain why certain reasoning or spatial leaps fail in the embedding geometry
OpenSSH 10.5 Release NotesThe OpenSSH team's frustration with AI models generating invalid security reports empirically supports the argument that LLMs have fundamental limitations — they can't reliably reason about complex security state machines
Emergent Introspective Awareness in Large Language ModelsBoth papers probe fundamental cognitive limitations of LLMs — 'LLMs can't jump' examines reasoning/generalization gaps, while this paper investigates whether self-reporting accurately reflects internal states, together building a picture of where LLM cognition is and isn't reliable.
Related
What Sort of Maths Are LLMs Good At?'LLMs can't jump' likely examines limitations of LLMs in structured reasoning tasks, directly complementing the new item's investigation of mathematical strengths and weaknesses
Compression is predictionBoth probe fundamental limitations of LLMs — 'LLMs can't jump' examines what models fail to generalize, while compression-as-prediction reveals what LLMs are actually doing when they model sequences, making them complementary perspectives on LLM capabilities
Challenged by
Claude Writes macOS Driver for Windows-Only HP PrinterThe piece argues LLMs have fundamental limitations on novel technical problems; generating a working macOS driver for obscure hardware represents exactly the kind of non-interpolative, domain-specific task that challenges this claim
Improving Gpt 5 6 Sol In Chatgpt'LLMs can't jump' argues LLMs face fundamental reasoning limitations; OpenAI's work on improving GPT problem-solving capabilities directly challenges or tests whether these limitations can be overcome through iterative improvement
DiffusionGemma Technical Report'LLMs can't jump' likely critiques limitations of sequential autoregressive generation; DiffusionGemma's parallel diffusion decoding represents an architectural response that fundamentally changes how tokens are generated, potentially addressing such limitations
Nova3D: Code-Native Generation of Programmable 3D Assets'LLMs can't jump' critiques LLM limitations in spatial/structured reasoning; Nova3D's success generating valid, hierarchical, constraint-laden Blender code challenges the notion that LLMs cannot handle complex structured 3D representations
mattwood.fyi