“The remaining thirteen it got wrong in all three attempts.”
Our reading, the sentence that was checked:
In the authors' 17 part-specification experiments, ChatGPT with DALL-E3 got 13 wrong in all three attempts.
Assessment: Contradicted Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Don’t Ride This Bike! Generative AI’s persistent trouble with compositionality and parts, paragraph 8.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-dont-ride-this-bike-generative-ais-306395-a1
We recorded 10 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 2024-12-08. 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
- informal experiments run November 10, 2024 (qualifier, stated by the speaker)
From the speaker
- 17 experiments
- The remaining thirteen
- wrong in all three attempts
- November 10, 2024
Our interpretation
- This is an exact count of experimental cases, not individual images.
- The count must be consistent with the linked experiment record's accepted outcomes.
- A case accepted as successful within three attempts cannot also count as failing all three.
Counts as supported if: The complete record identifies exactly thirteen of seventeen cases with no successful attempt within the allowed three attempts.
Counts as contradicted if: The complete record establishes a different count, including at least five cases accepted as successful within three attempts.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
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
The linked record accepts five cases within three attempts, including the glass table on open umbrellas. Five accepted cases out of seventeen leave at most twelve failures, not thirteen. This contradicts the exact tally under the experiment record's own scoring.
Limitations
- This is a counting finding, not an independent visual assessment of every output.
- The live record is mutable, and its annotation history was not established.
- The airplane is counted using the authors' accepted interpretation of wings.
Evidence (1 item)
Goes against the claim: Ernest Davis, NYU-hosted experimental record,
Correct
Why it matters here: The evaluator accepts the umbrella table in addition to the deer, wooden-legged table, cat, and gold airplane. Those five accepted cases leave at most twelve failures among seventeen.
- Publisher
- Ernest Davis, NYU-hosted experimental record
- Kind of source
- Demonstration
- Arithmetic
- Accepted cases = 3 + 1 + 1 = 5; maximum cases failing throughout = 17 - 5 = 12. (inputs: 17 experimental cases in the record; First-attempt successes: two-headed deer, glass table with wooden legs, glass table on open umbrellas; Second-attempt success: cat with human hands; Third-attempt success: gold airplane with purple wings, under the record's stated interpretation; Exact integer counts; no rounding.)
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Don’t Ride This Bike! Generative AI’s persistent trouble with compositionality and parts
The remaining thirteen it got wrong in all three attempts.
- Where
- paragraph 8
- Published
- 2024-12-08
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-dont-ride-this-bike-generative-ais-306395-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-dont-ride-this-bike-generative-ais-306395-a1.