“When we do so, we amplify the hype around AI, and lead people into thinking that these machines are trustworthy oracles capable of manipulation or decision-making, which they are not.”
Our reading, the sentence that was checked:
When we treat AI models like people, we amplify the hype around AI and lead people into thinking these machines are trustworthy oracles capable of manipulation or decision-making, which they are not.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Stop Treating AI Models Like People, paragraph 7.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-stop-treating-ai-models-like-people-29e613-a2
We recorded 9 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 2023-04-17. 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
- when we treat AI models like people (qualifier, stated by the speaker)
From the speaker
- When we do so
- amplify the hype around AI
- lead people into thinking
- trustworthy oracles capable of manipulation or decision-making
- which they are not
Our interpretation
- When we do so refers to treating AI models like people. This is a proposed causal relationship, not a conditional forecast.
- The ordinary reading of hype and trustworthy oracles is inflated expectations and unwarranted reliance on chatbot answers.
- The surrounding discussion of fabricated biographies supports reading the final clause as denying dependable oracle-like authority, rather than denying every capacity to influence people or produce a decision.
- The causal assertion concerns the chatbot setting described in early 2023. Later studies may inform the mechanism but do not automatically establish effects in that earlier population.
Counts as supported if: Relevant controlled comparisons show that anthropomorphic treatment increases inflated expectations or unwarranted trust, and contemporary evaluations establish that the targeted chatbots are insufficiently reliable for the authority attributed to them.
Counts as contradicted if: Relevant comparisons establish no increase or a reduction in inflated expectations and unwarranted trust, or demonstrate the dependable oracle-like reliability denied by the 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 reported on March 14, 2023 that GPT-4 fabricated facts and made reasoning errors, supporting the reliability concern [e5]. A study reported on May 9, 2024 found that adding voice increased perceived answer accuracy, while neither voice nor grammatical person affected overall system trust [e6]. These items do not establish or refute the claimed causal increase in hype and unwarranted reliance in the early-2023 chatbot setting [e5, e6].
Limitations
- The reliability disclosure concerns GPT-4 at release and does not test effects of anthropomorphic treatment.
- The later study's collection date and applicability to April 2023 are not established.
- Perceived accuracy and overall trust do not establish actual unwarranted reliance or public hype.
- This assessment reads the final clause as denying dependable oracle-like authority; it does not assess a denial of all manipulation or decision-making capacity.
Evidence (2 items)
Supports the claim: OpenAI,
not fully reliable
Why it matters here: This supports the limited reliability clause for GPT-4 at release. It does not demonstrate that anthropomorphic treatment causes hype or unwarranted trust.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: arXiv; Michelle Cohn and colleagues,
no effects of Modality
Why it matters here: The May 2024 report finds no effect of its manipulated cues on overall system trust, although voice increased perceived answer accuracy and some pronoun effects depended on context. These reported results qualify a broad trust claim; the collection date and applicability to April 2023 were not established.
- Publisher
- arXiv; Michelle Cohn and colleagues
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
Controlled comparisons applicable to early-2023 chatbot versions and users, holding response quality constant while varying anthropomorphic versus neutral treatment, and measuring inflated expectations and actual reliance against demonstrated reliability.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Stop Treating AI Models Like People
When we do so, we amplify the hype around AI, and lead people into thinking that these machines are trustworthy oracles capable of manipulation or decision-making, which they are not.
- Where
- paragraph 7
- Published
- 2023-04-17
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-stop-treating-ai-models-like-people-29e613-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-stop-treating-ai-models-like-people-29e613-a2.