“ChatGPT hasn’t done the conversion correctly (answer below), but the will to believe in this great new tool, full of precision (down to the second!), but lacking in reliability, is so strong that the author didn’t bother to check.”
Our reading, the sentence that was checked:
ChatGPT got a pace unit conversion wrong and lacks reliability, though it looks precise.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Scientists, please don’t let your chatbots grow up to be co-authors, paragraph 7.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-scientists-please-dont-let-your-chatbots-dfbd25-a1
We recorded 15 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 2023-01-14. 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)
From the speaker
- ChatGPT hasn’t done the conversion correctly
- full of precision (down to the second!)
- lacking in reliability
Our interpretation
- Assess the supplied normalized proposition about the conversion in Ehrsam’s January 13 tweet.
- Correctness means converting 4 minutes 21 seconds per kilometre to time per international mile, rounded to the nearest second.
- The reliability claim means that this apparently precise answer could not safely be accepted without checking. One example does not establish an overall error rate.
- The assertion about Ehrsam’s private checking and motivation is not established merely by his posting the answer.
Counts as supported if: The recorded answer differs from the correctly calculated pace beyond rounding, while presenting a seconds-level answer that Ehrsam publicly praised.
Counts as contradicted if: The recorded answer agrees with the correct conversion at its stated precision, or the recovered prompt materially differs from the one Marcus described.
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
The reported answer was 6:52 per mile. The independent conversion gives 7:00 to the nearest second. Ehrsam’s original tweet praised the answer’s precision. This supports the normalized observation that an apparently precise answer was wrong. It does not establish Marcus’s explanation of Ehrsam’s motives or a general ChatGPT error rate.
Limitations
- The numerical response is preserved in a contemporary personal-site account, not a readable original transcript or a reputable news report.
- Support applies to the conversion and localized reliability claim, not the psychological explanation.
Evidence (3 items)
Supports the claim: Az,
a 6:52 min/mile.
Why it matters here: This contemporary transcription identifies the erroneous answer. Applying the NIST definitions gives 420.038784 seconds per mile, which rounds to 7:00 rather than 6:52.
- Publisher
- Az
- Kind of source
- Secondary summary
- Arithmetic
- (4 × 60 + 21) × 5280 × 0.3048 / 1000 = 420.038784 seconds per mile (inputs: Reported input pace: 4 minutes 21 seconds per kilometre.; NIST section B.6: an international mile equals 5280 international feet.; NIST section B.6: an international foot equals 0.3048 metres.; Nearest whole second: 420 seconds, or 7 minutes 00 seconds.)
- Describes events of
- Public from
Supports the claim: Fred Ehrsam / Twitter,
ChatGPT delivers a precise and concise answer.
Why it matters here: The original tweet confirms that Ehrsam publicly praised the answer’s precision. It does not establish what checks he privately attempted.
- Where
- Tweet body
- Publisher
- Fred Ehrsam / Twitter
- Kind of source
- Demonstration
- Describes events of
- Public from
Context only: National Institute of Standards and Technology,
international foot
Why it matters here: The section records the international foot definition and the 5280-foot international mile. These supply the conversion constants; they are not observations of ChatGPT.
- Publisher
- National Institute of Standards and Technology
- Kind of source
- Official statistics
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Scientists, please don’t let your chatbots grow up to be co-authors
ChatGPT hasn’t done the conversion correctly (answer below), but the will to believe in this great new tool, full of precision (down to the second!), but lacking in reliability, is so strong that the author didn’t bother to check.
- Where
- paragraph 7
- Published
- 2023-01-14
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-scientists-please-dont-let-your-chatbots-dfbd25-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-scientists-please-dont-let-your-chatbots-dfbd25-a1.