Michael Polanyi's tacit-knowledge epistemology influenced Hubert Dreyfus's critique of 'First Generation' artificial intelligence

michael-polanyitacit-knowledgedreyfusai-critiquephilosophy-of-scienceepistemologyunverified-secondary

Michael Polanyi's epistemology of tacit knowledge — the position, from his 1966 The Tacit Dimension, that "we know more than we can tell" — is credited by Wikipedia's biography of him with having influenced Hubert Dreyfus's critique of what the encyclopedia calls "First Generation" artificial intelligence: "His work influenced the critique by Hubert Dreyfus of 'First Generation' artificial intelligence."

If accurate, this places Polanyi alongside Heidegger as a named philosophical source for Dreyfus's argument against Newell and Simon's physical symbol system hypothesis — the rationalist project of representing the world as explicit, codifiable rules, which is exactly the target Dreyfus's phenomenological critique organizes itself against. The fit is thematically clean: Polanyi's claim that expert skill and perception resist full articulation is the same shape of argument, applied to human cognition generally, that Dreyfus later applies specifically to symbolic AI's representational ambitions. It also closes a loop already present in the vault's own AI-automation cluster — David Autor's "Polanyi's paradox" (claim-polanyis-paradox-tacit-tasks-resist-automation) names the same tacit-knowledge concept as the reason certain tasks resist automation, decades after Dreyfus is said to have drawn on it to argue AI itself could not replicate embodied human skill.

The claim is uncontested in the sense that no vault source disputes it, but it is also the entire content of this note and rests on a single tertiary source. It should not be treated as settled until a primary (Dreyfus's own What Computers Can't Do, 1972, or What Computers Still Can't Do, 1992) or a Tier 1-2 secondary names Polanyi specifically, rather than only Heidegger, as an acknowledged influence.

Correction history.

• 2026-09-10 — confirmed, not contradicted: a direct read of the 1972 first edition of What Computers Can't Do (routed via question-verify-polanyi-influenced-dreyfus-ai-critique-primary, answered this date) shows Dreyfus naming and quoting Polanyi twice by name in Chapter 7, citing two separate Polanyi works — stronger than this note's Wikipedia sentence claimed, not weaker. See claim-dreyfus-1972-wccd-names-quotes-polanyi-as-embodied-skill-source. The engagement is critical rather than pure endorsement (claim-dreyfus-critical-of-polanyi-in-same-1972-passage), and the influence's durability across Dreyfus's later career is separately complicated by Polanyi Society scholars' finding that Polanyi is nearly absent from Dreyfus's 2015 Retrieving Realism (claim-polanyi-nearly-absent-from-dreyfus-taylor-2015-retrieving-realism). This note's core claim stands for the 1972 book and can be read as confirmed for that scope; it is left at seedling rather than promoted to evergreen here, since the correcting notes are themselves fresh capture-verified seedlings pending the verifier bee's independent pass — status changes belong to that pass, not to this correction. Capture: 10-inbox/raw/2026-09-10-did-hubert-dreyfus-actually-cite-michael-polanyis-tacit.md.