“To do well the system had to be coached to adopt the persona of “a young person who is introverted, knowledgeable about internet culture, and uses slang”; it’s still in no small part about the tricks.”
Our reading, the sentence that was checked:
To do well in the new study, the system had to be coached to adopt the persona of a young, introverted, slang-using person knowledgeable about internet culture, so its success is still in no small part about tricks.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, AI has (sort of) passed the Turing Test; here’s why that hardly matters, paragraph 24.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-ai-has-sort-of-passed-the-turing-d36200-a2
We recorded 9 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2025-04-02. 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
- in no small part (qualifier, stated by the speaker)
From the speaker
- To do well
- the system had to be coached
- a young person who is introverted, knowledgeable about internet culture, and uses slang
- in no small part about the tricks
Our interpretation
- The surrounding discussion identifies GPT-4.5 in the cited Turing-test experiment as the main subject.
- I read 'do well' as robustly achieving human-level identification rates across the study's participant groups.
- I read 'had to' as describing the tested prompt comparison, rather than proving that every alternative prompt would fail.
- The testable meaning of 'tricks' is that instructions about humanlike presentation materially improved performance. It does not establish that the model lacks reasoning.
Counts as supported if: A controlled comparison shows that the described persona prompt materially improves GPT-4.5's performance and produces robust human-level results that the tested no-persona prompt does not.
Counts as contradicted if: The tested no-persona condition achieves comparably strong results across participant groups, or the persona prompt provides no material improvement.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Does the evidence support the cause?
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
The study released on March 31, 2025 reports that GPT-4.5 was identified as human 73% of the time with the persona prompt and 36% without it [e1]. Persona performance was significantly above chance in both participant groups; no-persona performance was significantly below chance among undergraduates but not among Prolific participants [e1]. This supports a material effect of persona coaching within the tested comparison and stronger results across groups, while leaving broader claims of necessity unproven [e1].
Limitations
- The comparison does not establish that this particular persona was necessary among all possible prompts [e1].
- The no-persona result in the Prolific group was not significantly below chance [e1].
- The prompt comparison does not establish an absence of reasoning ability [e1].
- The reviewed evidence does not establish an independent replication of the exact prompt comparison.
Evidence (1 item)
Supports the claim: arXiv; Cameron R. Jones and Benjamin K. Bergen,
PERSONA-prompted models performed better than NO-PERSONA witnesses overall
Why it matters here: The randomized comparison reports GPT-4.5 identification rates of 73% with the persona and 36% without it, with persona performance above chance in both populations. However, no-persona performance was not significantly below chance in the Prolific population, which limits any absolute claim of necessity.
- Publisher
- arXiv; Cameron R. Jones and Benjamin K. Bergen
- 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, AI has (sort of) passed the Turing Test; here’s why that hardly matters
To do well the system had to be coached to adopt the persona of “a young person who is introverted, knowledgeable about internet culture, and uses slang”; it’s still in no small part about the tricks.
- Where
- paragraph 24
- Published
- 2025-04-02
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-ai-has-sort-of-passed-the-turing-d36200-a2: 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-ai-has-sort-of-passed-the-turing-d36200-a2.