“What is new that they have at last become widely recognized, not just as the transitory “bugs” the industry wanted you to think of, but as inherent limitations that flow from the very design of LLMs they really are.”
Our reading, the sentence that was checked:
LLMs' technical problems are now widely recognized as inherent limitations of their design, not transitory bugs.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, The AI bubble is all over now, baby blue, paragraph 9.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-the-ai-bubble-is-all-over-now-baby-fc11e1-a1
We recorded 7 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 2025-12-26. 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
- have at last become widely recognized
- not just as the transitory “bugs”
- inherent limitations
- flow from the very design of LLMs
Our interpretation
- The ordinary reading is broad recognition among informed researchers and industry participants that important reliability problems arise from how LLMs work, rather than being isolated implementation defects.
- Widely recognized does not require unanimity, but a few selected statements cannot establish it.
- The claim also implies a recent change in recognition. That requires comparison with earlier views.
- Use the 2025 prepublication research and industry-discussion period because no narrower period or regular reporting cadence is named.
- The surrounding rhetoric does not by itself establish that inherent means impossible to mitigate under every possible system design.
Counts as supported if: Representative surveys or a broad, systematically assembled record show widespread acceptance of design-based limitations by publication, with evidence that such recognition had recently expanded.
Counts as contradicted if: Comparable evidence shows that design-based explanations remained a minority view, or that the claimed broad recognition was already established well before the period described as new.
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 record contains explicit architectural explanations and industry acknowledgment of structural causes. It also contains qualifications and proposed mitigations. These sources establish that the view existed and was publicly discussed, but do not measure how widespread it was or whether widespread recognition was new in late 2025.
Limitations
- A selection of technical papers is not a representative opinion survey.
- Understanding why errors arise differs from accepting that they cannot be remedied.
- The relevant audience and meaning of widespread remain unspecified.
Evidence (3 items)
Context only: arXiv; Richard Ackermann and Simeon Emanuilov,
a structural outcome
Why it matters here: This directly expresses the view Marcus describes. One preprint cannot establish widespread acceptance or a recent shift in opinion.
- Where
- Abstract
- Publisher
- arXiv; Richard Ackermann and Simeon Emanuilov
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: OpenAI,
abstain when uncertain
Why it matters here: OpenAI discusses statistical causes in training and evaluation, but explicitly rejects inevitable hallucination because models can abstain. Recognition of structural causes therefore cannot automatically be counted as agreement with an unfixable-limit interpretation.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: arXiv; Adam Tauman Kalai and Santosh S. Vempala,
may mitigate
Why it matters here: This earlier paper derives a hallucination bound for particular facts under a calibration condition, while distinguishing other errors that may be mitigated. It establishes prior structural analysis, not prior widespread agreement.
- Publisher
- arXiv; Adam Tauman Kalai and Santosh S. Vempala
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
Representative, comparable surveys or a systematic record of researcher and industry positions before and during 2025, directly distinguishing temporary defects, structural limitations, and unavoidable failures.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, The AI bubble is all over now, baby blue
What is new that they have at last become widely recognized, not just as the transitory “bugs” the industry wanted you to think of, but as inherent limitations that flow from the very design of LLMs they really are.
- Where
- paragraph 9
- Published
- 2025-12-26
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-the-ai-bubble-is-all-over-now-baby-fc11e1-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-ai-bubble-is-all-over-now-baby-fc11e1-a1.