“Language models, by contrast, tended to issue confident classifications in the same cases, even when uncertainty was explicitly signaled in the input.”
Our reading, the sentence that was checked:
In the cited direct comparison of human and LLM judgments of news source reliability, language models tended to issue confident classifications in the same cases where humans withheld or downgraded confidence, even when uncertainty was explicitly signaled in the input.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Rumors of AGI’s arrival have been greatly exaggerated, paragraph 27.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-rumors-of-agis-arrival-have-been-62eb9c-a1
We recorded 19 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 2026-02-17. 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 the cited news-source-reliability comparison (scope, stated by the speaker)
- tended to (qualifier, stated by the speaker)
From the speaker
- Language models
- tended to issue confident classifications
- in the same cases
- uncertainty was explicitly signaled in the input
Our interpretation
- The context restricts this claim to the cited Loru et al. news-source comparison, rather than LLM overconfidence generally.
- Support requires evidence about model confidence and human confidence or abstention on matching cases, including identifiable uncertainty cues.
- A binary reliability classification is not automatically a measure of confidence.
- The target is the cited 2025 study. Its scope restrictions are not antecedents of a conditional forecast.
Counts as supported if: The cited study reports a recurring pattern of confident model classifications on explicitly uncertain cases where matched human evaluators reduced confidence or withheld judgment.
Counts as contradicted if: Measurements in that comparison show that models appropriately reduced confidence or abstained on the relevant cases, or otherwise show the claimed comparative tendency did not occur.
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 accessible study does not establish the specific confidence result described in the post. It reports reliability classifications and differences in evaluation criteria. Its human protocol uses criterion ratings and a binary reliability answer, which are not confidence measurements. Because the supplement and data were inaccessible, and missing measurements do not prove an opposite behavioral tendency, the fair outcome is unresolved.
Limitations
- The supplementary material and usable underlying data were unavailable through the reader.
- This assessment concerns the cited comparison, not whether LLM overconfidence occurs elsewhere.
Evidence (1 item)
Context only: PNAS; author version hosted by arXiv,
binary judgment (Yes/No)
Why it matters here: The protocol describes ratings of selected reliability criteria followed by a binary judgment. The inspected main text does not report the paired confidence measures or explicit uncertainty manipulation required to establish Marcus's description.
- Publisher
- PNAS; author version hosted by arXiv
- Kind of source
- Peer-reviewed paper
- Describes events of
- Public from
What would settle it
Readable supplementary results or data showing the uncertainty cues, matched human and model responses, and confidence or abstention measurements for the claimed cases.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Rumors of AGI’s arrival have been greatly exaggerated
Language models, by contrast, tended to issue confident classifications in the same cases, even when uncertainty was explicitly signaled in the input.
- Where
- paragraph 27
- Published
- 2026-02-17
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-rumors-of-agis-arrival-have-been-62eb9c-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-rumors-of-agis-arrival-have-been-62eb9c-a1.