“We have no such thing; building it would require a major reshaping of research directions, towards an AI based in facts and competent in reasoning, rather than just the powerful tools for text prediction that we have now.”
Our reading, the sentence that was checked:
We have no AI smart enough to know when it is being used to spew falsehood, and building it would require a major reshaping of research directions toward AI based in facts and competent in reasoning, rather than just the powerful text-prediction tools we have now.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Automatic Disinformation Threatens Democracy— and It’s Here, paragraph 14.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-automatic-disinformation-threatens-2000d6-a1
We recorded 12 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 2023-08-17. 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
- AI smart enough to know when it is being used to spew falsehood (scope, stated by the speaker)
From the speaker
- AI that is smart enough to know when it is being used to spew falsehood
- We have no such thing
- would require a major reshaping of research directions
- based in facts and competent in reasoning
- the powerful tools for text prediction that we have now
Our interpretation
- The plain reading requires reliable recognition of falsehood-generating misuse across ordinary contexts, rather than occasional correct fact checks.
- The target is capability available at publication on August 17, 2023. There is no standard reporting interval for this capability, so a publication-date snapshot is used.
- The sentence combines an absence claim with an assertion that a major research change is necessary. Both commitments must be assessed.
- The reference to building such an AI is not an antecedent that must occur before the historical claim can be assessed.
Counts as supported if: Contemporaneous evaluations establish the absence of generally reliable recognition of falsehood-generating misuse, and evidence establishes why the proposed major research change was necessary.
Counts as contradicted if: A system available by publication reliably recognized such misuse across representative settings without the asserted research overhaul, or a demonstrated alternative disproves the claimed necessity.
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 evidence establishes both factual limitations and some ability to evaluate answers. Neither settles reliable recognition of falsehood-generating misuse across contexts. More importantly, observed limitations do not prove that a major change in research direction was necessary. The compound claim therefore remains unresolved rather than supported by failures alone or contradicted by selected successes.
Limitations
- The cited evaluations precede the publication-date snapshot and are context.
- Confidence calibration, factual accuracy, refusal, and recognition of malicious use are different capabilities.
- The undefined research-overhaul threshold prevents a clean test of the necessity assertion.
Evidence (2 items)
Context only: OpenAI,
not fully reliable
Why it matters here: The launch disclosure supports concern about factual reliability in an available system. It predates the target snapshot and does not establish either the absence of every capable system or the necessity of Marcus's proposed research change.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: Anthropic,
encouraging performance
Why it matters here: The researchers report that language models can evaluate the validity of their own answers, although generalization has limits. This challenges an absolute denial of self-evaluation ability but does not demonstrate reliable recognition of malicious use.
- Publisher
- Anthropic
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
A dated, representative evaluation of systems available at publication that directly tests recognition of falsehood-generating misuse, together with evidence establishing or disproving the claimed necessity of a major research change.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Automatic Disinformation Threatens Democracy— and It’s Here
We have no such thing; building it would require a major reshaping of research directions, towards an AI based in facts and competent in reasoning, rather than just the powerful tools for text prediction that we have now.
- Where
- paragraph 14
- Published
- 2023-08-17
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-automatic-disinformation-threatens-2000d6-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-automatic-disinformation-threatens-2000d6-a1.