“without even acknowledging, for example, the massive problem LLMs have with fabricating citations”
Our reading, the sentence that was checked:
LLMs have a massive problem with fabricating citations.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, OpenAI’s o3 and Tyler Cowen’s Misguided AGI Fantasy, paragraph 5.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-openais-o3-and-tyler-cowens-misguided-80c113-a1
We recorded 4 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 2025-04-17. 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)
From the speaker
- LLMs
- massive problem
- fabricating citations
Our interpretation
- The plain reading is that fabricated references are a substantial, recurring reliability problem across commonly used LLM systems. It does not mean every model fabricates most citations.
- Because this describes an ongoing state, the target period is early 2025 through publication, rather than earlier generations or subsequent improvements.
- The assessment concerns citation fabrication, not the separate allegation that Cowen failed to acknowledge it.
Counts as supported if: Contemporaneous evaluations document substantial, recurring fabrication of references or source links across multiple commonly used LLM systems.
Counts as contradicted if: Broad evaluations of contemporary systems establish that fabricated references are rare across relevant citation tasks, undermining the claimed scale of the problem.
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
The contemporary audit supports the ordinary reading that invented citations were a serious, recurring problem. It found fabricated links across several systems and particularly high combined fabricated-or-broken-link rates for Gemini and Grok 3. This supports the qualitative warning, not a universal failure rate or the claim that most LLM citations were fabricated.
Limitations
- Massive is a qualitative judgment.
- The study tested a specific news-retrieval task and found substantial differences between systems.
- Its combined broken-link measure must not be presented as a fabrication-only percentage.
Evidence (1 item)
Supports the claim: Tow Center for Digital Journalism / Columbia Journalism Review,
fabricated or broken URLs
Why it matters here: The researchers documented invented source links across multiple systems, supporting a substantial citation-reliability problem. Their reported majority rate combines fabricated and broken URLs, so it is not a measured fabrication-only rate.
- Publisher
- Tow Center for Digital Journalism / Columbia Journalism Review
- Kind of source
- Independent measurement
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, OpenAI’s o3 and Tyler Cowen’s Misguided AGI Fantasy
without even acknowledging, for example, the massive problem LLMs have with fabricating citations
- Where
- paragraph 5
- Published
- 2025-04-17
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-openais-o3-and-tyler-cowens-misguided-80c113-a1: 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.
About assessment marcus-openais-o3-and-tyler-cowens-misguided-80c113-a1.