“This is one attempt to articulate that, and as the paper notes, current techniques capture only a fraction of what ultimately needs to be captured:”
Our reading, the sentence that was checked:
As the paper notes, current techniques capture only a fraction of the many facets of cognition that ultimately need to be captured.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Is AGI the right goal for AI?, paragraph 16.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-is-agi-the-right-goal-for-ai-5bdbd2-a1
We recorded 11 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2025-10-16. 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
- as the paper notes (qualifier, stated by the speaker)
From the speaker
- as the paper notes
- current techniques
- capture only a fraction
- what ultimately needs to be captured
Our interpretation
- The referent is the cognitive framework in A Definition of AGI, rather than an independently established inventory of everything intelligence requires.
- The ordinary reading is substantial incompleteness across that framework's cognitive abilities. Fraction does not specify a percentage and need not mean less than half.
- Current refers to the October 2025 research snapshot available when the post appeared. The target period uses that reporting month; later improvements do not answer the claim.
- As the paper notes is an attribution, not a conditional forecast.
- Because Marcus coauthored the paper, its assertions identify his framework but cannot independently establish that his substantive claim is correct.
Counts as supported if: Independent assessments of the contemporaneous systems across the framework's cognitive domains establish substantial gaps in multiple required abilities.
Counts as contradicted if: Independent assessments show that contemporaneous systems broadly met the framework's requirements, with only minor remaining gaps.
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 referenced manuscript makes the incompleteness argument Marcus attributes to it and reports partial aggregate scores. However, he coauthored it, so those assertions cannot independently establish that he was right. An independent visual benchmark documents a specific gap, but does not settle coverage across the full cognitive framework. The searched evidence therefore does not settle the substantive claim under the required source standard.
Limitations
- This is an evidence gap, not evidence that the cognitive gaps were absent.
- The framework's weighting and the meaning of fraction affect how the statement is read.
- Neither a benchmark percentage nor an aggregate score directly measures a percentage of human cognition.
Evidence (2 items)
Context only: arXiv; Dan Hendrycks and coauthors,
narrower than well-educated humans
Why it matters here: The original arXiv version reports a ten-domain framework and estimated scores of 27% for GPT-4 and 58% for GPT-5. This establishes what the framework asserts, but Marcus's coauthorship prevents treating it as independent confirmation; the arXiv deposit also postdates his post.
- Publisher
- arXiv; Dan Hendrycks and coauthors
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: ClockBench; Alek Safar,
limitations of current frontier LLMs.
Why it matters here: This independent study reports a large human–model gap in recognizing valid times on unusual analog clocks. It supplies evidence of a specific visual limitation, but neither measures the entire cognitive framework nor establishes October performance.
- Where
- Abstract
- Publisher
- ClockBench; Alek Safar
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
An independent, reproducible assessment of publication-time model versions across the framework's domains, with transparent tasks and scoring, would establish whether the remaining gaps were substantial.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Is AGI the right goal for AI?
This is one attempt to articulate that, and as the paper notes, current techniques capture only a fraction of what ultimately needs to be captured:
- Where
- paragraph 16
- Published
- 2025-10-16
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-is-agi-the-right-goal-for-ai-5bdbd2-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-is-agi-the-right-goal-for-ai-5bdbd2-a1.