“And of course Generative AI still has problems with common sense, as an endless of stream of examples have shown, this one involving circulating on X today:”
Our reading, the sentence that was checked:
Generative AI still has problems with common sense.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Is AI already killing people by accident?, paragraph 6.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-is-ai-already-killing-people-by-accident-c3d575-a1
We recorded 14 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 2026-03-01. 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
- Generative AI still has problems with common sense
- an endless stream of examples
- this one circulating on X today
Our interpretation
- Plain reading: as of the post, widely used generative AI models still often failed simple common-sense questions that humans answer easily.
- The example 'circulating on X today' is most likely the viral February 2026 car-wash question, but the claim does not depend on that identification.
- Target period: 2026-01-01 to 2026-03-01, since he points to then-current examples.
Counts as supported if: Documented, reproducible common-sense failures by then-current mainstream models in early 2026.
Counts as contradicted if: Systematic tests in early 2026 show then-current models reliably passing such common-sense questions.
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
In February 2026, a simple common-sense question went viral: should you walk or drive to a car wash 50 metres away when you want to wash your car? A systematic test of 53 models found only 5 answered correctly every time, and 33 never did. News coverage of the viral thread found many mainstream models from OpenAI, Google, Anthropic and xAI failing. Some top models passed, and humans in a 10,000-person baseline also missed it about 28.5% of the time. Still, these documented, reproducible failures support the plain claim that generative AI 'still has problems with common sense'.
Limitations
- The main systematic test comes from a commercial vendor.
- Some frontier models passed reliably, so the claim holds for the field broadly, not for every model.
Evidence (3 items)
Supports the claim: Opper,
Out of 53 models, only 5 can do this reliably. 15 more can sometimes get there but unpredictably. The remaining 33 never get it right.
Why it matters here: Most current models failed a trivial common-sense question, repeatedly, in February 2026.
- Publisher
- Opper
- Kind of source
- Independent measurement
- Describes events of
- Public from
Supports the claim: GIGAZINE,
However, only the GPT-5.2 Thinking, Claude Opus 4.6, and Gemini 3 Pro models answered this test correctly.
Why it matters here: Confirms the failures were widely circulating on X in mid-February 2026, across many mainstream models.
- Where
- paragraph 4
- Publisher
- GIGAZINE
- Kind of source
- News report
- Describes events of
- Public from
Context only: Opper,
Turns out GPT-5 (7/10) answered about as reliably as the average human (71.5%) in this test.
Why it matters here: Humans also missed this question about 28.5% of the time, and a few top models passed reliably. This tempers, but does not negate, the finding.
- Where
- Part 3
- Publisher
- Opper
- Kind of source
- Independent measurement
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Is AI already killing people by accident?
And of course Generative AI still has problems with common sense, as an endless of stream of examples have shown, this one involving circulating on X today:
- Where
- paragraph 6
- Published
- 2026-03-01
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-is-ai-already-killing-people-by-accident-c3d575-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-is-ai-already-killing-people-by-accident-c3d575-a1.