“Over the last couple days I have been asking version 2 to draw images identifying and tagging parts of images. The results are consistently awful:”
Our reading, the sentence that was checked:
When Marcus asked Grok 2 over the last couple of days to draw images identifying and tagging parts of images, the results were consistently awful.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Elon Musk’s terrifying vision for AI, paragraph 14.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-elon-musks-terrifying-vision-for-120653-a1
We recorded 9 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 2025-02-16. 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
- his own requests over the last couple of days (scope, stated by the speaker)
- drawing images identifying and tagging parts of images (qualifier, stated by the speaker)
From the speaker
- Over the last couple days
- I have been asking version 2
- draw images identifying and tagging parts of images
- The results are consistently awful
Our interpretation
- The target is Marcus’s own set of requests, not Grok 2’s general performance.
- I interpret 'awful' as material errors in identifying parts, rendering their labels, or attaching labels to the correct parts.
- I interpret 'consistently' as a recurring pattern across his runs, rather than an isolated selected failure.
- The approximate period is February 14–16, 2025, allowing for tests on the posting day.
Counts as supported if: An authenticated, sufficiently complete record of his prompts and outputs shows recurring material labeling errors across the stated testing period.
Counts as contradicted if: An authenticated record of the same runs shows generally correct labeling, with failures too isolated to justify the claimed consistency.
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 claim is checkable as a report about a particular set of image-labeling tests. Searches did not locate independent evidence sufficient to inspect that set or establish consistent failure. The missing record does not show that Marcus was wrong.
Limitations
- Text extraction cannot establish what the embedded images show.
- Later tests or general benchmarks would not establish what happened in his particular runs.
Evidence (0 items)
No evidence item settled this claim by the cutoff.
What would settle it
Authenticated prompts and outputs from the stated days, with timestamps and the full run count, plus independent inspection of labeling accuracy.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Elon Musk’s terrifying vision for AI
Over the last couple days I have been asking version 2 to draw images identifying and tagging parts of images. The results are consistently awful:
- Where
- paragraph 14
- Published
- 2025-02-16
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-elon-musks-terrifying-vision-for-120653-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-elon-musks-terrifying-vision-for-120653-a1.