“By the standards articulated in the original definitions of artificial general intelligence—robustness across environments, reliable generalization under novelty, and autonomous goal-directed behavior—current AI systems remain limited.”
Our reading, the sentence that was checked:
By original AGI standards, current AI systems remain limited.
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 30.
- 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-c64d40-a1
We recorded 19 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
- by original AGI standards (qualifier, stated by the speaker)
From the speaker
- by the standards articulated in the original definitions of AGI
- robustness across environments
- reliable generalization under novelty
- autonomous goal-directed behavior
- current AI systems remain limited
Our interpretation
- Plain reading: in February 2026, frontier AI systems clearly fell short of human-level robustness, generalization to novel situations, and autonomous goal-directed behavior.
- 'Current AI systems' means the frontier systems available at the post date (for example Claude Opus 4.6, released February 2026). The target period runs from February 2026 through March 2026, so that it covers the first independent evaluations of those same systems. This choice is flagged as an ambiguity.
- 'Remain limited' is a weak claim. It is supported if public evaluations of those systems show a large gap from human performance on tests built for novel environments and self-directed goal discovery.
Counts as supported if: Independent evaluations of frontier systems from the period show large shortfalls against humans on tasks requiring generalization to novel environments, inferring goals without instructions, or robust autonomous behavior.
Counts as contradicted if: Independent evaluations from the period show frontier systems matching humans on novel-environment, goal-discovery and robustness tests without task-specific 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 that, judged by robustness, generalization to new situations and autonomous goal-directed behavior, AI systems at the time still fell short. ARC-AGI-3 was released in March 2026. It drops agents into unfamiliar environments with no instructions and no stated goals. Frontier models scored under 1%, including Claude Opus 4.6, which was available when the post appeared. Humans solved all the environments. That is a large, measured gap on exactly the abilities the claim names. Some experts argued in Nature that human-level AI had already arrived, but that is an opinion piece, not a measurement. 'Remain limited' is a modest claim, and the evidence supports it.
Limitations
- The decisive benchmark results came out about five weeks after the post.
- ARC-AGI-3 is one benchmark, and its scoring heavily penalizes inefficiency.
- 'Original AGI definitions' are paraphrased by the speaker.
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: On a benchmark built to test generalization in novel environments, frontier systems from the period (including Claude Opus 4.6, released before the post) scored under 1%. Humans solved every environment. This is a large shortfall on the speaker's named standards.
- Publisher
- Decrypt
- Kind of source
- News report
- Describes events of
- Public from
Supports the claim: Decrypt,
drop an AI agent into an unfamiliar game-like world with zero instructions, zero stated goals, and no description of the rules. The agent has to explore, figure out what it's supposed to do, form a plan, and execute it.
Why it matters here: Shows the test targets the same abilities the claim names: generalization under novelty and autonomous goal-directed behavior.
- Publisher
- Decrypt
- 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: Some experts argued at the time that human-level general intelligence had already arrived. This is an opinion piece, 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: Primary description of the benchmark's design, which matches the claim's criteria.
- 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
By the standards articulated in the original definitions of artificial general intelligence—robustness across environments, reliable generalization under novelty, and autonomous goal-directed behavior—current AI systems remain limited.
- Where
- paragraph 30
- Published
- 2026-02-17
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-rumors-of-agis-arrival-have-been-c64d40-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-rumors-of-agis-arrival-have-been-c64d40-a1.