“As I have long argued, capturing this without cognitive models and common sense is really hard.”
Our reading, the sentence that was checked:
Capturing the pragmatic implications of language without cognitive models and common sense is really hard.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, What “game over” for the latest paradigm in AI might look like, paragraph 15.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-what-game-over-for-the-latest-paradigm-e1c19a-a1
We recorded 13 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 2022-10-29. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- 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
- without cognitive models and common sense (qualifier, stated by the speaker)
From the speaker
- capturing this
- without cognitive models and common sense
- is really hard
Our interpretation
- This refers to reliably interpreting pragmatic implications, particularly the conversational implicatures illustrated in the surrounding passage.
- The ordinary reading is substantial practical difficulty for contemporary systems lacking the stated resources. It does not assert impossibility.
- Lacking an explicit symbolic component is not automatically equivalent to lacking learned cognitive representations or commonsense knowledge.
- For this ongoing observation, the target is the latest evaluation period reported in the study specifically identified by the post. Its API experiment timestamps place the relevant testing in August–September 2022.
Counts as supported if: Evaluations show substantial, persistent difficulty interpreting pragmatic implications in contemporary systems whose lack of the relevant cognitive and commonsense resources is independently established.
Counts as contradicted if: Systems established to lack those resources reliably interpret a suitably broad range of pragmatic implications without substantial difficulty.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
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
The cited study supports the narrower observation that pragmatic interpretation challenged contemporary language models. Its strongest prompted model also reached what the researchers called near-human performance. Neither result establishes whether the systems lacked cognitive representations or commonsense knowledge. Scoring the full statement would require silently equating the absence of explicit machinery with the absence of those resources. The evidence therefore leaves the full claim unresolved.
Limitations
- The finding does not deny the demonstrated difficulty.
- A near-human result on one binary task does not establish reliable pragmatic understanding generally.
- This observation is assessed against contemporary systems, not later models.
Evidence (2 items)
Context only: Laura Ruis and colleagues, arXiv,
still show a significant gap with human performance.
Why it matters here: The original study establishes difficulty even for the better-performing adapted models; its API timestamps place their evaluations in August. It does not independently establish that these models lacked cognitive representations or commonsense knowledge.
- Publisher
- Laura Ruis and colleagues, arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: Laura Ruis and coauthors,
performance jumps to 80.6% for text-davinci-002, which is near-human performance!
Why it matters here: The researchers describe their strongest few-shot result as near-human, qualifying any broad suggestion of uniformly poor performance. The experiment date comes from Table 59 of the original paper; this result still does not establish which cognitive or commonsense resources the model possessed.
- Publisher
- Laura Ruis and coauthors
- Kind of source
- Independent measurement
- Describes events of
- Public from
What would settle it
Contemporary evaluations that independently characterize the systems' cognitive and commonsense resources and test reliable pragmatic interpretation across varied tasks and prompting conditions.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, What “game over” for the latest paradigm in AI might look like
As I have long argued, capturing this without cognitive models and common sense is really hard.
- Where
- paragraph 15
- Published
- 2022-10-29
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-what-game-over-for-the-latest-paradigm-e1c19a-a1: Current. Outcome: Unresolved. 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-what-game-over-for-the-latest-paradigm-e1c19a-a1.