“Systems like DALL-E 2, GPT-3, Flamingo, and Gato appear to be mind-blowing, but nobody who has looked at them carefully would confuse them for human intelligence.”
Our reading, the sentence that was checked:
Systems like DALL-E 2, GPT-3, Flamingo and Gato appear mind-blowing, but nobody who has looked at them carefully would confuse them for human intelligence.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, The New Science of Alt Intelligence, paragraph 17.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-the-new-science-of-alt-intelligence-75a6f1-a1
We recorded 20 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-29. The source may have been edited after 2022-05-14. 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
- who has looked at them carefully (qualifier, stated by the speaker)
From the speaker
- DALL-E 2, GPT-3, Flamingo, and Gato
- nobody who has looked at them carefully
- would confuse them for human intelligence
Our interpretation
- The ordinary contextual reading concerns broad, flexible human intelligence, rather than human-level performance on a particular task.
- The statement also asserts agreement among careful observers; demonstrations of errors do not independently establish that agreement.
- I use April 1 through May 14, 2022, as the latest public evaluation cycle surrounding the systems discussed. This is an ongoing observation, not a claim about subsequent versions.
- Mind-blowing is an expression of admiration and is not separately scored.
Counts as supported if: Careful contemporaneous evaluations covering all four systems establish a clear distinction from broad human intelligence and substantiate the asserted agreement among informed observers.
Counts as contradicted if: A documented careful contemporary evaluator regards a named system as exhibiting broadly equivalent human intelligence, or a suitable evaluation establishes such equivalence.
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 document important limitations, and a contemporary GPT-3 analysis also credits meaningful semantic abilities. Neither finding settles Marcus's broader assertion about human intelligence or what every careful observer would conclude. Scoring it as supported would require replacing the asserted agreement with the narrower claim that the systems made revealing errors.
Limitations
- The literal and rhetorical readings of nobody require different evidence.
- The quoted GPT-3 evidence comes from the article's readable abstract.
- Task-specific failures are not a complete measure of human intelligence.
Evidence (2 items)
Context only: arXiv; DeepMind,
examples of such behaviours
Why it matters here: The preceding discussion identifies hallucinations and ungrounded guesses in Flamingo's open-ended dialogue. These establish limitations, but neither a comprehensive human comparison nor unanimity among careful observers.
- Publisher
- arXiv; DeepMind
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: Minds and Machines; Springer Nature,
truth-telling is not amongst them
Why it matters here: The authors distinguish GPT-3's semantic competence from its unreliable truth-telling and challenge wholesale dismissals of its semantic abilities. This qualifies a strong reading of Marcus's dismissal without establishing broad human equivalence.
- Where
- Abstract
- Publisher
- Minds and Machines; Springer Nature
- Kind of source
- Peer-reviewed paper
- Describes events of
- Public from
What would settle it
Contemporaneous evaluations and documented observer judgments addressing broad human equivalence across the named systems under an explicit, defensible comparison.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, The New Science of Alt Intelligence
Systems like DALL-E 2, GPT-3, Flamingo, and Gato appear to be mind-blowing, but nobody who has looked at them carefully would confuse them for human intelligence.
- Where
- paragraph 17
- Published
- 2022-05-14
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-the-new-science-of-alt-intelligence-75a6f1-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-the-new-science-of-alt-intelligence-75a6f1-a1.