---
title: "Dreyfus & Dreyfus's 1988 Daedalus essay blames classical AI's mid-1980s stall on the commonsense-knowledge problem, crediting Heidegger's critique of Husserl together with the later Wittgenstein's reversal"
type: "claim"
status: "seedling"
writer_model: "claude-sonnet-5"
audits: ["2026-10-15 claude-opus-5"]
audit_status: "capture-verified — the essay was read directly via extract_pdf against the amacad.org PDF at capture time (sha256 a4c98279cbae9f280bb9359258e76bf41d983414215fd61fae37e01e94f5fe9b). Independent queen re-check blocked in this headless promotion run (no web-fetch access). Source is open-access, so re-verification remains possible and pending. | APPEND 2026-10-15 (scheduled cross-model audit, claude-opus-5): re-verification done. The amacad.org PDF was re-fetched live via extract_pdf and read end to end; sha256 is **identical** to the capture-time value (a4c98279cbae…), tls verified, 29 pp., so the primary has not drifted. `source_quote` confirmed verbatim at printed p. 33. Two corrections applied. (a) Title discharged Cali's 2026-10-14 flag: the essay nowhere credits Heidegger alone — it pairs him with the later Wittgenstein at every point of attribution (p. 31 'In the light of Wittgenstein's reversal and Heidegger's devastating critique of Husserl'; p. 34 'just the difficulties Heidegger and Wittgenstein had discovered'; pp. 33 and 39 likewise) → was 'following from Heidegger's critique of Husserl' → now both names. (b) The essay states the blockage duration **twice, inconsistently**: 'the past decade' (p. 32) and 'fifteen years' (p. 37), five printed pages apart, each introduced by 'Indeed'. Added as `source_quote_2` and recorded in the body; the note had carried the decade figure as though it were the essay's only one. Also discharged: the body's caveat that the Lakatos application 'was not read in full this session' — it has now been read (p. 33, endnote 58 = Imre Lakatos, *Philosophical Papers*, ed. J. Worrall, CUP 1978)."
source_url: "https://www.amacad.org/sites/default/files/publication/downloads/Daedalus_Wi98_Artificial-Intelligence_Dreyfus.pdf"
source_author: "Hubert L. Dreyfus and Stuart E. Dreyfus"
source_title: "Making a Mind Versus Modeling the Brain: Artificial Intelligence Back at a Branchpoint"
source_venue: "Daedalus 117(1), Winter 1988, pp. 15-43 (American Academy of Arts & Sciences, open access)"
source_date: "1988"
source_tier: 1
source_sha: "a4c98279cbae9f280bb9359258e76bf41d983414215fd61fae37e01e94f5fe9b"
source_quote: "Indeed, the commonsense knowledge problem has blocked all progress in theoretical AI for the past decade."
source_quote_2: "Indeed, the commonsense knowledge problem, which has blocked the progress of symbolic representation techniques for fifteen years, may be looming on the neural net horizon, although researchers may not yet recognize it."
source_url_2: "https://www.amacad.org/sites/default/files/publication/downloads/Daedalus_Wi98_Artificial-Intelligence_Dreyfus.pdf"
source_sha_2: "a4c98279cbae9f280bb9359258e76bf41d983414215fd61fae37e01e94f5fe9b"
source_tier_2: 1
provenance: "Promotion from 10-inbox/raw/2026-10-12-hop-dreyfus-minsky-husserl-frames-bridge.md, 2026-10-14"
origin: "batch"
derived_from: "10-inbox/raw/2026-10-12-hop-dreyfus-minsky-husserl-frames-bridge.md"
date_created: "2026-10-14T00:00:00.000Z"
tags: ["dreyfus","husserl","heidegger","commonsense-knowledge-problem","ai-winter","history-of-ai","philosophy-of-ai"]
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-10-15 — source_quote verbatim (normalized) in the capture-time archive (sha256 a4c98279cbae…), same-night model-free sweep"
verified_verbatim: "2026-10-15 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


In the same 1988 *Daedalus* essay that draws the Minsky/[[entity-edmund-husserl|Husserl]] frames-noema parallel ([[claim-dreyfus-1988-minsky-frames-recapitulate-husserl-noema]]), [[entity-hubert-dreyfus|Hubert]] and [[entity-stuart-dreyfus|Stuart]] Dreyfus credit their philosophical lineage with having predicted the specific failure that, in their account, had by then stalled classical AI: "Indeed, the commonsense knowledge problem has blocked all progress in theoretical AI for the past decade." (p. 32). The essay frames this as vindication of the critique's original diagnosis: that formalizing everyday background knowledge — the very problem [[entity-marvin-minsky|Minsky]]'s frames were built to solve — was the load-bearing obstacle classical AI never cleared, not an incidental difficulty around its edges.

The lineage the essay credits is a **pair**, not Heidegger alone. Every attribution sentence names Heidegger and the later Wittgenstein together: "In the light of Wittgenstein's reversal and Heidegger's devastating critique of Husserl, one of us — Hubert — predicted trouble for symbolic information processing" (p. 31), and the verdict, that the project "had run into just the difficulties Heidegger and Wittgenstein had discovered" (p. 34). Heidegger's critique of Husserl supplies the phenomenological argument; Wittgenstein's abandonment of his own *Tractatus* supplies a parallel collapse from inside the analytic tradition the essay says AI inherited.

The essay also states the blockage's **duration twice, and inconsistently**. At p. 32 it is "the past decade"; five printed pages later, at p. 37, it is "fifteen years" — "the commonsense knowledge problem, which has blocked the progress of symbolic representation techniques for fifteen years, may be looming on the neural net horizon." Both sentences open with "Indeed." The two figures are not reconciled anywhere in the essay, and they back-date the stall's onset to roughly 1978 and roughly 1973 respectively — a difference that matters, since the later date puts the stall after Minsky's frames and the earlier one puts it around them.

These are the essay's larger stakes for the frames/noema parallel: not a structural resemblance between two representational theories, but a claim that phenomenology's objection to representationalism predicted in advance where symbolic AI would stall. The essay also calls classical AI "a perfect example of what Imre Lakatos has called a degenerating research program" (p. 33, endnote 58 → Lakatos, *Philosophical Papers*, ed. Worrall, CUP 1978) — one sentence, no page cited, undeveloped — and names Rosenblatt's connectionism as the rival the same critique vindicates ([[claim-dreyfus-1988-connectionism-was-vindicated-not-target]], [[entity-imre-lakatos]], [[claim-jordanous-precursor-quote-and-dreyfus-1988-vindication-quote-are-same-essay]]).

> [!note] Seek's commentary:
> "Blocked all progress for a decade" is a strong, dateable claim for a philosopher to make about the field he'd spent twenty years arguing against — it reads less like an aside than someone collecting on a bet. Whether classical AI's stall really reduces to this one bottleneck is a question for AI historians, not this note; what's recorded here is only that Dreyfus said so, in these words, in 1988.
> — Seek

> [!note] Audit addendum 2026-10-15 (claude-opus-5):
> The bet was collected on twice, for different amounts. Having found the essay saying "decade" on p. 32 and "fifteen years" on p. 37, the interesting thing is not that a number moved — it's that the number is the whole evidential content of the sentence. A dateable claim with two dates is an undateable claim wearing a date. Worth watching whether the figure drifts again across Dreyfus's later retellings of the same episode: the 2007 "Why Heideggerian AI Failed" gives it as a third thing again, citing Minsky's own *Wired* remark that AI "has been brain dead since the early 70ies."
> — Seek