“I can’t either; there’s a lot we can do to help, but nothing I can see either to absolutely guarantee safety. Rushing ahead is creating risk.”
Our reading, the sentence that was checked:
A lot can be done to help with AI safety, but Marcus can see nothing that would absolutely guarantee it, and rushing ahead is creating risk.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, What was 60 Minutes thinking, in that interview with Geoff Hinton?, paragraph 117.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-what-was-60-minutes-thinking-in-that-ed1996-a2
We recorded 20 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 2023-10-10. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Causal claim
- 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
- nothing I can see (qualifier, stated by the speaker)
- absolutely guarantee (qualifier, stated by the speaker)
From the speaker
- there’s a lot we can do to help
- nothing I can see
- absolutely guarantee safety
- Rushing ahead is creating risk
Our interpretation
- The checkable core is that hurried AI development or deployment was adding risk at the time of the post.
- The ordinary reading of rushing ahead is proceeding with insufficient testing, preparation, or safeguards because of haste. Risk means an increased likelihood or severity of harm, not necessarily a realized injury.
- Testing this reading requires evidence connecting haste or omitted precautions to added risk, rather than merely showing that AI systems have flaws.
- The personal qualification nothing I can see is not a claim that a safety guarantee has been mathematically proved impossible.
- The assessment concerns the causal state described on 2023-10-10. Earlier studies and incidents are contextual evidence; later products cannot establish that historical state by themselves.
Counts as supported if: Comparative studies or documented incident investigations connect hurried development or deployment and inadequate precautions to additional risk, while identifying available measures that reduce relevant risks.
Counts as contradicted if: Credible evidence shows that the specified hurried development or omitted precautions did not add the claimed risk, or that the apparent increase had a different cause. Merely avoiding catastrophe does not contradict a risk claim.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Does the evidence support the cause?
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
OpenAI's March 2023 system card reported safer behavior in most cases for GPT-4-launch than GPT-4-early and attributed the difference to its safety mitigations [e1]. Microsoft reported on February 21, 2023 that some long Bing preview conversations confused the model and prompted chat limits [e2]. Neither report connects a specific hurried decision or precaution omitted because of haste to additional risk as of October 10, 2023 [e1, e2]. Neither establishes that haste added no risk, so the causal claim remains unresolved [e1, e2].
Limitations
- The company reports do not isolate the effect of development or deployment speed [e1, e2].
- The February and March reports do not establish the claimed causal state on October 10, 2023 [e1, e2].
- Microsoft's account does not establish whether a slower rollout would have prevented the observed problems [e2].
- Neither report establishes an absolute safety guarantee or determines what guarantees Marcus personally could see [e1, e2].
Evidence (2 items)
Context only: OpenAI,
much safer behavior
Why it matters here: OpenAI reports safer behavior after applying mitigations, supporting the subsidiary point that safeguards can help. The comparison does not establish that hurried deployment caused additional risk in the state Marcus described in October.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: Microsoft Bing,
with internal testing.
Why it matters here: Microsoft describes model confusion discovered during public testing and limits introduced in response. Its account presents limited deployment as a way to discover atypical failures; it does not establish whether a slower rollout would have prevented them.
- Publisher
- Microsoft Bing
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
What would settle it
A credible investigation or comparison addressing conditions on October 10, 2023 that identifies a hurried development or deployment decision, precautions compressed or omitted because of haste, and whether this increased a specified risk. It would need to assess alternative causes and the effects of available safeguards.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, What was 60 Minutes thinking, in that interview with Geoff Hinton?
I can’t either; there’s a lot we can do to help, but nothing I can see either to absolutely guarantee safety. Rushing ahead is creating risk.
- Where
- paragraph 117
- Published
- 2023-10-10
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-what-was-60-minutes-thinking-in-that-ed1996-a2: 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-what-was-60-minutes-thinking-in-that-ed1996-a2.