“Automation is a double-edged sword, and there is a kind of uncanny valley. We know perfectly well not to trust lousy systems; and wouldn’t need to pay attention to truly reliable systems, but the closer they get to perfect, the easier it is for mere mortals to space out.”
Our reading, the sentence that was checked:
As automated systems approach but do not reach perfection, humans overseeing them become more likely to stop paying attention.
Assessment: Mixed Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, The CNET Fake News Fiasco, Autopilot, and the Uncanny Cognitive Valley, paragraph 25.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-the-cnet-fake-news-fiasco-autopilot-3a4f96-a1
We recorded 10 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 2023-01-31. 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
- systems that are good but not truly reliable (qualifier, stated by the speaker)
From the speaker
- the closer they get to perfect
- the easier it is for mere mortals to space out
- wouldn’t need to pay attention to truly reliable systems
Our interpretation
- The testable claim is that increasing the reliability of imperfect automation tends to reduce attentive human oversight.
- Space out includes reduced monitoring and increased task-unrelated mind wandering; these are related but distinct outcomes.
- The statement is read as a general tendency, not an assertion that every reliability increase affects every operator.
- The imperfect-system qualification defines the causal comparison; it is not an antecedent to a conditional forecast.
Counts as supported if: Controlled comparisons of otherwise similar systems show poorer monitoring, reduced attention, or increased mind wandering with higher imperfect reliability.
Counts as contradicted if: Controlled comparisons show unchanged or improved attention as reliability rises. Differences across tasks or attention measures can warrant a mixed assessment rather than a universal conclusion.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Does the evidence support the cause?
Outcome: Mixed
Part of the claim holds and part does not, and the record says which part is which.
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
Higher reliability worsened monitoring in a controlled cockpit simulation, supporting Marcus's practical warning. A separate controlled autopilot study did not find that higher reliability increased task-unrelated mind wandering. The monitoring-performance component holds in the first setting; the broader suggestion that increasing reliability makes people space out is not consistent across the reviewed measures and settings.
Limitations
- The null study had only 16 participants and does not prove absence of an effect in other settings.
- The supporting study was assessed through its accessible author abstract.
- These experiments do not establish a universal or steadily increasing effect near perfect reliability.
Evidence (2 items)
Supports the claim: Human Factors and Ergonomics Society; record hosted by University of Central Florida,
poorer monitoring performance
Why it matters here: The experiment compared 50% and 90% reliability during simulated cockpit multitasking and found poorer monitoring with greater reliability. The availability date conservatively uses the university record's stated repository date.
- Publisher
- Human Factors and Ergonomics Society; record hosted by University of Central Florida
- Kind of source
- Peer-reviewed paper
- Describes events of
- Public from
Goes against the claim: Frontiers in Human Neuroscience,
Attentional states were not correlated
Why it matters here: Sixteen participants supervised an autopilot under two reliability conditions lasting 45 minutes each. Reliability changed trust and perceived demand, but the study did not find the predicted increase in task-unrelated mind wandering, limiting the broad causal claim.
- Publisher
- Frontiers in Human Neuroscience
- Kind of source
- Peer-reviewed paper
- Describes events of
- Public from
What would settle it
Replicated comparisons spanning high and near-perfect reliability, with workload controlled and both monitoring behavior and mind wandering measured, would establish the effect's scope.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, The CNET Fake News Fiasco, Autopilot, and the Uncanny Cognitive Valley
Automation is a double-edged sword, and there is a kind of uncanny valley. We know perfectly well not to trust lousy systems; and wouldn’t need to pay attention to truly reliable systems, but the closer they get to perfect, the easier it is for mere mortals to space out.
- Where
- paragraph 25
- Published
- 2023-01-31
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-the-cnet-fake-news-fiasco-autopilot-3a4f96-a1: Current. Outcome: Mixed. 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-the-cnet-fake-news-fiasco-autopilot-3a4f96-a1.