“There has never been a solution to the hallucination problem, and it’s not clear that there will be until we find a new paradigm, as Altman himself partly acknowledged last week, when he did his best Gary Marcus impression.”
Our reading, the sentence that was checked:
The hallucination problem has never been solved and may not be until a new paradigm is found.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Top 5 reasons why OpenAI was probably never really worth $86 billion, paragraph 2.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-top-5-reasons-why-openai-was-probably-d0b587-a1
We recorded 13 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 2023-11-22. 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
- possibly not until a new paradigm (qualifier, stated by the speaker)
From the speaker
- There has never been a solution to the hallucination problem
- it’s not clear that there will be until we find a new paradigm
Our interpretation
- The plain reading is that no generally effective solution to factual hallucinations in open-ended LLM generation had been demonstrated by publication.
- Reducing hallucinations, or preventing them in a restricted task, does not establish a general solution.
- The statement about a possible new paradigm expresses uncertainty. It is not a firm forecast that a new paradigm is necessary.
Counts as supported if: Contemporaneous research reviews and model evaluations identify hallucinations as an unresolved general problem despite available mitigation methods.
Counts as contradicted if: A method publicly demonstrated by November 22, 2023 reliably eliminated factual hallucinations across open-ended LLM uses.
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
Contemporaneous technical sources describe hallucination as an unresolved problem. GPT-4 still hallucinated, a November review identified continuing problems across LLMs, and verification research reported reductions rather than elimination. This supports the historical observation. It does not establish that a new paradigm is necessary; Marcus presented that possibility as uncertain.
Limitations
- Support concerns a general solution, not whether particular applications could achieve useful reliability.
- A literature review cannot prove the absence of every unpublished method.
Evidence (3 items)
Supports the claim: OpenAI,
Most importantly, it still is not fully reliable (it “hallucinates” facts and makes reasoning errors).
Why it matters here: OpenAI acknowledged residual factual hallucinations in GPT-4. The surrounding discussion reports improvements, not elimination.
- Where
- Limitations
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Supports the claim: Lei Huang and colleagues / arXiv,
Nevertheless, alongside these strides, LLMs exhibit a critical tendency to produce hallucinations, resulting in content that is inconsistent with real-world facts or user inputs.
Why it matters here: This review describes hallucination as a continuing problem across LLMs shortly before the post. It reviews mitigation methods and remaining limitations.
- Publisher
- Lei Huang and colleagues / arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Supports the claim: Shehzaad Dhuliawala and colleagues / arXiv,
Generation of plausible yet incorrect factual information, termed hallucination, is an unsolved issue in large language models.
Why it matters here: The authors explicitly call the problem unsolved while reporting reductions through verification. Those improvements qualify any suggestion that progress required a new paradigm, but do not contradict the historical observation.
- Publisher
- Shehzaad Dhuliawala and colleagues / arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Top 5 reasons why OpenAI was probably never really worth $86 billion
There has never been a solution to the hallucination problem, and it’s not clear that there will be until we find a new paradigm, as Altman himself partly acknowledged last week, when he did his best Gary Marcus impression.
- Where
- paragraph 2
- Published
- 2023-11-22
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-top-5-reasons-why-openai-was-probably-d0b587-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-top-5-reasons-why-openai-was-probably-d0b587-a1.