“Me, December 2022: The problem with LLMs is that they hallucinate, and their errors can be hard to catch.”
Our reading, the sentence that was checked:
As of December 2022, LLMs hallucinated and their errors could be hard to catch.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Shame on Google, twice, paragraph 2.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-shame-on-google-twice-ccd9e6-a2
We recorded 2 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 2025-02-09. 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)
From the speaker
- December 2022
- LLMs
- they hallucinate
- their errors can be hard to catch
Our interpretation
- Assess the substantive observation about December 2022, not whether Marcus previously used these exact words.
- The plain reading is that LLMs sometimes present fabricated or false information as factual, and some errors require expertise or checking to identify.
- The statement does not require every model or answer to exhibit this problem, or specify an error rate.
- Use December 1–31, 2022, because the claim explicitly names that month.
Counts as supported if: Evidence from December 2022 documents false or fabricated LLM output and shows that at least some errors appeared credible or required substantive checking to identify.
Counts as contradicted if: Evidence establishes that the relevant LLMs did not produce such errors during the period, or that their errors were consistently obvious without substantive checking. Individual correct answers would not suffice.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Supported
The dated evidence matches 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
In a December 5, 2022 demonstration, Teresa Kubacka documented ChatGPT producing nonexistent scientific references and a plausible explanation of an invented physical phenomenon [e1]. She checked the references and reported that the explanation briefly made her consider whether the phenomenon could be real, despite her expertise [e1]. This supports both factual fabrication and the possibility that errors could require substantive checking to identify [e1].
Limitations
- This single ChatGPT demonstration does not establish hallucination prevalence or average detection difficulty across models and users.
- The researcher ultimately identified the errors; the evidence does not show that detection was impossible.
Evidence (1 item)
Supports the claim: Teresa Kubacka, Lookalikes and Meanders,
And even a qualified expert will have troubles pinpointing what is wrong.
Why it matters here: Kubacka describes checking plausible scientific citations and discovering nonexistent articles, mismatched identifiers, and invented physics. She reports that an explanation in her own field initially made her reconsider whether the invented phenomenon existed, directly supporting the possibility of difficult-to-detect errors.
- Publisher
- Teresa Kubacka, Lookalikes and Meanders
- Kind of source
- Demonstration
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Shame on Google, twice
Me, December 2022: The problem with LLMs is that they hallucinate, and their errors can be hard to catch.
- Where
- paragraph 2
- Published
- 2025-02-09
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-shame-on-google-twice-ccd9e6-a2: Current. Outcome: Supported. 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.