“Current systems simply do not postulate such cognitive models—and so, they technically can’t prove themselves wrong, because they do not even have a representation of what being “right” would look like.”
Our reading, the sentence that was checked:
Current systems do not postulate cognitive models, so they technically cannot prove themselves wrong, because they have no representation of what being right would look like.
Assessment: Contradicted Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Three ideas from linguistics that everyone in AI should know, paragraph 34.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-three-ideas-from-linguistics-that-accdff-a1
We recorded 26 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-29. The source may have been edited after 2022-06-22. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Causal claim
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
Conditions
- technically (qualifier, stated by the speaker)
- current systems (scope, stated by the speaker)
From the speaker
- Current systems
- do not postulate such cognitive models
- they technically can’t prove themselves wrong
- they do not even have a representation of what being “right” would look like
Our interpretation
- Current systems means language-model systems existing by June 22, 2022, including fine-tuned systems.
- The ordinary functional reading of prove themselves wrong is identifying a substantive error in their own output using relevant facts or task constraints, with the criticism independently checkable.
- This does not require a formal mathematical proof or spontaneous correction without prompting.
- The claim links the alleged absence of cognitive representations to an absolute inability to establish errors in the system's own answers.
Counts as supported if: Architecture and controlled evaluations establish both the asserted absence of relevant representations and the resulting inability to identify and substantiate errors in the system's own outputs.
Counts as contradicted if: A system available before the post produces independently assessable critiques of its own errors and uses them to improve its answers. That contradicts the claimed absolute inability without proving a particular cognitive architecture.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Does the evidence support the cause?
Outcome: Contradicted
The dated evidence goes against what the claim committed to, under the criteria frozen before grading.
Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.
- Evidence as of
- Deadline state when assessed
- No horizon
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
A paper public before the post reports language models critiquing their own answers and using their own critiques to improve them. That contradicts the absolute inability under the ordinary functional reading. The evidence concerns trained research systems and bounded tasks. It does not establish formal self-proof, universal reliability or an explicit cognitive world model.
Limitations
- The result depends on treating independently assessable self-critique as demonstrating an error.
- The models received task-specific training and prompts.
- Human evaluation of summarization errors was imperfect.
Evidence (1 item)
Goes against the claim: arXiv; William Saunders and colleagues, OpenAI,
refining their own summaries
Why it matters here: The experiments evaluate models critiquing their own outputs against supplied passages or task constraints, and using critiques to improve answers. This contradicts the functional claim of complete inability, while leaving the nature of internal cognitive representations unsettled.
- Publisher
- arXiv; William Saunders and colleagues, OpenAI
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Three ideas from linguistics that everyone in AI should know
Current systems simply do not postulate such cognitive models—and so, they technically can’t prove themselves wrong, because they do not even have a representation of what being “right” would look like.
- Where
- paragraph 34
- Published
- 2022-06-22
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-three-ideas-from-linguistics-that-accdff-a1: Current. Outcome: Contradicted. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.
About assessment marcus-three-ideas-from-linguistics-that-accdff-a1.