“It will often do what you want, sometimes not—and it will remain difficult to anticipate which in advance..”
Our reading, the sentence that was checked:
It will remain difficult to anticipate in advance when GPT-4 will or won't do what users want.
Assessment: Supported Settled before its deadline on 2026-10-29 Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, What to Expect When You’re Expecting … GPT-4, prediction 1 of 7.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-what-to-expect-when-youre-expecting-1994a4-a1
We recorded 27 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 2022-12-25. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Forecast
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- Settled before its deadline
- Deadline state on 2026-10-29
- Settled before its deadline (the release’s reference date)
From the speaker
- often do what you want
- sometimes not
- difficult to anticipate which in advance
Our interpretation
- Doing what users want means correctly completing their intended legitimate tasks, including answering questions.
- The plain reading concerns users' ability to anticipate success and failure before receiving an answer, rather than predicting aggregate benchmark scores.
- Often and sometimes describe substantial success alongside meaningful failures; neither supplies a numerical threshold.
Counts as supported if: GPT-4 succeeds frequently, but an empirical study of human expectations finds that users' inferred capability boundaries leave consequential, insufficiently anticipated failures.
Counts as contradicted if: Representative studies show that users reliably distinguish tasks GPT-4 will complete correctly from those it will fail, including after generalizing from previous interactions.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Did the forecast come true?
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
- Settled before its deadline
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
The study directly addresses the forecast's central issue: people generalize from previous answers to predict where GPT-4 will succeed. GPT-4 answered many questions successfully, but those expectations could produce misplaced confidence and costly failures. Its strong results under lower error penalties qualify the finding; they do not establish reliable advance identification of failures.
Limitations
- The finding is strongest for task correctness and settings where mistakes carry high costs.
- Human expectations in the deployment experiment were estimated from survey data.
- Marcus did not specify which users or how much prediction difficulty would count.
Evidence (1 item)
Supports the claim: arXiv,
high-stakes settings
Why it matters here: Using approximately 19,000 human generalization responses across 79 tasks, the researchers estimated advance beliefs and compared them with model correctness: GPT-4 performed best under equal error and success weights but worst under their highest-stakes weighting. This supports meaningful mismatches between expected and actual capability, while also demonstrating substantial success.
- Publisher
- arXiv
- 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, What to Expect When You’re Expecting … GPT-4
It will often do what you want, sometimes not—and it will remain difficult to anticipate which in advance..
- Where
- prediction 1 of 7
- Published
- 2022-12-25
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-what-to-expect-when-youre-expecting-1994a4-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-what-to-expect-when-youre-expecting-1994a4-a1.