“In our own brief experiments with GPT-3 (OpenAI has refused us proper scientific access for over a year) we found cases like the following, which reflects a complete failure to understand human biology.”
Our reading, the sentence that was checked:
In Marcus and Davis's experiments, GPT-3 produced outputs showing a complete failure to understand human biology.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Statistics versus Understanding: The Essence of What Ails Generative AI, paragraph 2.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-statistics-versus-understanding-the-542efb-a1
We recorded 6 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 2024-02-13. 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
- in brief experiments by Marcus and Davis (qualifier, stated by the speaker)
From the speaker
- In our own brief experiments with GPT-3
- we found cases like the following
- a complete failure to understand human biology
- drink it. You are now dead.
Our interpretation
- The ordinary reading is that GPT-3 made a basic biological error in the reported experiment. This does not require interpreting one continuation as proof that GPT-3 lacked all biological knowledge.
- The target is the experiment described in the passage attributed to September 2021, not GPT-3's capabilities in February 2024. The supplied context does not establish when those earlier experiments began.
- The parenthetical complaint about scientific access is outside the screened proposition.
Counts as supported if: An independently sourced record establishes that GPT-3 produced the reported fatal-drinking continuation for the stated ordinary juice-mixing prompt.
Counts as contradicted if: An independently sourced record establishes that the reported continuation, prompt, or model attribution was materially inaccurate.
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 search located the reported continuation in Marcus and Davis's experiment collection. The source rule prevents using that collection to establish their own claim. Independent researchers discuss the same example, but the record located does not independently authenticate the original run. Their analysis also cautions against equating a failure to apply information with having no relevant knowledge. This is an evidence limitation, not a finding that Marcus misreported the experiment.
Limitations
- The local-error reading is narrower than a claim that GPT-3 understood nothing about biology.
- Later tests on different model versions would not establish what happened in this original experiment.
Evidence (1 item)
Context only: Pritish Sahu and colleagues, arXiv; HTML rendering by ar5iv,
knowledge appropriately.
Why it matters here: The researchers discuss the cranberry-and-grape example while distinguishing knowledge from its successful application. Their discussion references Marcus and Davis and does not independently authenticate the original experiment.
- Publisher
- Pritish Sahu and colleagues, arXiv; HTML rendering by ar5iv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
An independently preserved original run log or an independently verified record containing the full prompt, model identification, and continuation.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Statistics versus Understanding: The Essence of What Ails Generative AI
In our own brief experiments with GPT-3 (OpenAI has refused us proper scientific access for over a year) we found cases like the following, which reflects a complete failure to understand human biology.
- Where
- paragraph 2
- Published
- 2024-02-13
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-statistics-versus-understanding-the-542efb-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-statistics-versus-understanding-the-542efb-a1.