โWe still donโt have a solid test of what o3 does without the pretraining, in the case that would be more comparable to humansโ
Our reading, the sentence that was checked:
There is no solid test yet of o3's ARC performance without pretraining.
Assessment: Supported 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-a18448-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
- We still donโt have a solid test
- what o3 does without the pretraining
- more comparable to humans
Our interpretation
- We have means the publicly available scientific record.
- A solid test means a documented evaluation of the December o3 system without ARC-specific training.
- The human-comparability phrase motivates the requested control; it does not assert that removing ARC training would equalize every aspect of human and model experience.
- The ongoing state is assessed over the latest relevant reporting interval: the December 20 announcement through the December 22 post.
Counts as supported if: The public evaluation record lacks an ARC-untrained result, and evaluator statements or contemporaneous scrutiny confirm that the effect of ARC-specific training remains unmeasured.
Counts as contradicted if: A reliable ARC evaluation of the relevant system without ARC-specific training was publicly available by December 22, 2024.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Supported
The dated evidence matches 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
The evaluator explicitly says it had not tested the ARC-untrained model. Contemporaneous independent commentary likewise treats performance without ARC training as unknown. No qualifying public result was found for the period. This supports Marcus's claim about missing public evidence, without establishing that OpenAI never ran a private experiment.
Limitations
- The evaluator page has later updates, and the training note's precise insertion date is not separately archived in this record.
- Removing ARC training alone would not make model and human testing conditions identical.
Evidence (2 items)
Supports the claim: ARC Prize,
not yet tested the ARC-untrained
Why it matters here: The evaluator explicitly reports the missing control. This directly supports the absence of a public evaluator result separating performance from ARC-specific training.
- Publisher
- ARC Prize
- Kind of source
- Independent measurement
- Describes events of
- Public from
Context only: Melanie Mitchell, AI: A Guide for Thinking Humans,
without this
Why it matters here: An independent researcher discussing the announcement also identifies performance without ARC training as unknown. This corroborates the public information gap without establishing what OpenAI tested privately.
- Where
- Footnote 5
- Publisher
- Melanie Mitchell, AI: A Guide for Thinking Humans
- Kind of source
- Secondary summary
- Describes events of
- Public from
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
We still donโt have a solid test of what o3 does without the pretraining, in the case that would be more comparable to humans
- Where
- paragraph 13
- Published
- 2024-12-22
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-c39-a18448-a1: Current. Outcome: Supported. 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.