“Current AI systems are powerful and increasingly useful tools, but they do not exhibit the flexible, self-directed competence that the original concept of artificial general intelligence was intended to capture.”
Our reading, the sentence that was checked:
Current AI systems are powerful and increasingly useful tools, but they do not exhibit the flexible, self-directed competence that the original concept of AGI was intended to capture.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Rumors of AGI’s arrival have been greatly exaggerated, paragraph 31.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-rumors-of-agis-arrival-have-been-520739-a2
We recorded 15 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2026-02-17. 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)
Conditions
- as defined by the original concept of AGI (qualifier, stated by the speaker)
From the speaker
- current AI systems are powerful and increasingly useful tools
- do not exhibit the flexible, self-directed competence
- that the original concept of AGI was intended to capture
Our interpretation
- Plain reading: in February 2026, frontier AI was useful and improving, but could not flexibly handle unfamiliar situations or set and pursue its own goals the way a human can.
- 'Current AI systems' means frontier systems available at the post (for example Claude Opus 4.6). The target period is February-March 2026, to include the first independent evaluations of those systems. This choice is flagged.
- The 'powerful and increasingly useful' part is a concession, not the main claim; the main claim is the lack of flexible, self-directed competence.
Counts as supported if: Evaluations from the period show frontier systems failing at tasks that require exploring unfamiliar settings and working out goals without instructions, where humans succeed, while also showing rising capability on narrower tasks.
Counts as contradicted if: Evaluations from the period show frontier systems matching humans at flexible, self-directed tasks in novel environments without human-built help.
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 claim is restricted to the original concept of AGI as the speaker describes it. Its main part is that AI systems at the time lacked flexible, self-directed competence. The ARC Prize Foundation says ARC-AGI-3 tests agents on exploring novel environments and acquiring goals on the fly (page dated 2026-03-25). Decrypt reported on 2026-03-26 that frontier models scored under 1% at launch: Gemini 3.1 Pro 0.37%, GPT-5.4 0.26%, Claude Opus 4.6 0.25%, Grok-4.20 0%. Humans solved 100% of environments. The concession that systems were powerful and increasingly useful also holds. OfficeChai reported on 2026-02-21 that METR measured Opus 4.6 at a 50% time horizon of about 14.5 hours on software, machine-learning and cybersecurity tasks, on a rising trend. A Nature comment dated 2026-02-02 argued that human-level AI had already arrived. It is argument, not measurement.
Limitations
- Most of the weight rests on one benchmark, ARC-AGI-3. Its 100% mark means matching human efficiency, not only solving the games.
- The ARC-AGI-3 model scores come from a news report, not the foundation's own results page.
- The ARC-AGI-3 results came out about five weeks after the post.
- The METR estimate has a wide confidence interval (6 to 98 hours), and METR said its task suite was nearly saturated.
- 'Original concept of AGI' is the speaker's own characterization, and 'flexible, self-directed competence' has no agreed measure.
Evidence (4 items)
Supports the claim: Decrypt,
Google’s Gemini 3.1 Pro led the pack at 0.37%. OpenAI’s GPT-5.4 came in at 0.26%. Anthropic’s Claude Opus 4.6 managed 0.25%, while xAI’s Grok-4.20 scored exactly zero. Humans, meanwhile, solved 100% of environments.
Why it matters here: Frontier systems scored under 1% where humans scored 100% on tasks requiring self-directed exploration and goal discovery. This is direct evidence of missing flexible, self-directed competence.
- Publisher
- Decrypt
- Kind of source
- News report
- Describes events of
- Public from
Supports the claim: OfficeChai,
METR’s latest measurements put Claude Opus 4.6 at a 50%-time horizon of approximately 14.5 hours
Why it matters here: Shows strong, rising capability on well-specified tasks. This is consistent with the claim's concession that systems are powerful and increasingly useful.
- Publisher
- OfficeChai
- Kind of source
- News report
- Describes events of
- Public from
Goes against the claim: Nature (Comment by Chen, Belkin, Bergen, Danks),
The vision of human-level machine intelligence laid out by Alan Turing in the 1950s is now a reality.
Why it matters here: An expert opinion that general intelligence had already been reached. It is an argument, not a measurement, so it is context only.
- Where
- standfirst
- Publisher
- Nature (Comment by Chen, Belkin, Bergen, Danks)
- Kind of source
- News report
- Describes events of
- Public from
Context only: ARC Prize Foundation,
challenges AI agents to explore novel environments, acquire goals on the fly, build adaptable world models, and learn continuously.
Why it matters here: Shows the benchmark measures flexible, self-directed competence of the kind the claim describes.
- Publisher
- ARC Prize Foundation
- Kind of source
- Benchmark result
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Rumors of AGI’s arrival have been greatly exaggerated
Current AI systems are powerful and increasingly useful tools, but they do not exhibit the flexible, self-directed competence that the original concept of artificial general intelligence was intended to capture.
- Where
- paragraph 31
- Published
- 2026-02-17
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-rumors-of-agis-arrival-have-been-520739-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-rumors-of-agis-arrival-have-been-520739-a2.