โWhat we saw is not AGI.โ
Our reading, the sentence that was checked:
o3 as demonstrated is not AGI.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-c39-333f47-a1
We recorded 17 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 2024-12-22. 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 releaseโs reference date)
From the speaker
- What we saw
- is not AGI
Our interpretation
- What we saw refers to the December 20 o3 preview, not subsequently released models.
- The ordinary reading of AGI is broad human-level cognitive capability, including learning and solving unfamiliar problems across domains.
- Marcus specifies no operational threshold. Neither a particular ARC score nor flawless performance on every easy task is imposed as the definition.
- The categorical assertion that the system was not AGI is stronger than saying the demonstration did not establish AGI.
Counts as supported if: Evidence about the demonstrated system establishes substantial limitations incompatible with broad human-level cognitive capability under the ordinary reading.
Counts as contradicted if: Reliable evaluations of that same system establish broad human-level cognitive capability across sufficiently varied unfamiliar tasks and domains.
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 records show strong but incomplete ARC performance and an evaluator applying a demanding AGI criterion. They justify caution about calling the demonstration AGI. They do not, by themselves, settle whether the system lacked broad human-level cognitive capability. Supporting the categorical claim would require adopting an additional definition or treating failure to demonstrate AGI as proof of its absence.
Limitations
- This assesses the categorical wording, not the narrower claim that ARC performance alone did not prove AGI.
- Expert agreement is informative but does not replace a defined capability test.
- Later released o3 systems are outside the target.
Evidence (2 items)
Context only: Franรงois Chollet, X thread reproduced by Nitter,
becomes outright impossible.
Why it matters here: The evaluator supplies a stringent criterion involving the inability to construct human-easy, AI-hard tests. Applying that criterion as dispositive would add a threshold that Marcus did not explicitly provide.
- Publisher
- Franรงois Chollet, X thread reproduced by Nitter
- Kind of source
- Benchmark result
- Describes events of
- Public from
Context only: Mikel Bober-Irizar,
366/400 tasks in the ARC-AGI-Pub dataset.
Why it matters here: The analysis concerns the December results and documents substantial success with remaining failures. Neither incomplete performance nor high performance on this one benchmark settles broad general intelligence.
- Where
- Opening paragraph
- Publisher
- Mikel Bober-Irizar
- Kind of source
- Independent measurement
- Describes events of
- Public from
What would settle it
An explicit operational interpretation of AGI and sufficiently broad evaluations of the December preview that establish whether it meets that interpretation.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, ๐ผ๐ฏ โ๐๐ฅ๐ ๐๐๐โ ๐ฝ๐ผ๐๐๐บ๐ผ๐ฟ๐๐ฒ๐บ ๐บ๐ฒ๐ด๐ฎ๐๐ต๐ฟ๐ฒ๐ฎ๐ฑ: ๐๐ต๐ ๐๐ต๐ถ๐ป๐ด๐ ๐ด๐ผ๐ ๐ต๐ฒ๐ฎ๐๐ฒ๐ฑ, ๐๐ต๐ฎ๐ ๐๐ฒ๐ป๐ ๐๐ฟ๐ผ๐ป๐ด, ๐ฎ๐ป๐ฑ ๐๐ต๐ฎ๐ ๐ถ๐ ๐ฎ๐น๐น ๐บ๐ฒ๐ฎ๐ป๐
What we saw is not AGI.
- Where
- paragraph 16
- Published
- 2024-12-22
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-c39-333f47-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.