“We aren’t close to AGI, but we are reshaping our world on the premise that we are.”
Our reading, the sentence that was checked:
We are not close to AGI.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, The Grand AGI Delusion, paragraph 1.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-the-grand-agi-delusion-2aaae2-a1
We recorded 2 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 2025-10-12. 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
- We aren’t close to AGI
Our interpretation
- The ordinary reading is that AI remains substantially short of broadly human-level general cognitive competence.
- Marcus does not define AGI, specify a capability threshold, or say how many years would count as close.
- I use 2025 through the post date as the reporting period for this ongoing field-wide assessment. This is an evaluator-selected year-to-date window, not a deadline supplied by Marcus.
- Failure to demonstrate AGI does not by itself establish that AGI is distant. Individual benchmark successes do not establish proximity either.
Counts as supported if: Broad, contemporary capability evidence establishes a substantial remaining gap to general human-level competence, with a defensible connection between that gap and the ordinary meaning of not close.
Counts as contradicted if: A system available during the target period demonstrably achieves general human-level competence, or broad evidence establishes proximity using a defensible measure of the remaining gap.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
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 evidence documents substantial capability gaps alongside strong performance in demanding scientific tasks. It does not establish how far those systems were from AGI. Marcus's statement concerns proximity, which is stronger than saying AGI had not been demonstrated. Scoring it either way would require a definition of close that he did not provide and the cited evaluations do not supply.
Limitations
- Benchmark failures do not measure time remaining to AGI.
- Specialized successes do not demonstrate broad general competence.
- The astronomy study is a preprint, and the interactive benchmark was preliminary.
Evidence (4 items)
Context only: ARC Prize Foundation,
single-digit percentage scores.
Why it matters here: The launch evaluation identifies a substantial gap on unfamiliar reasoning tasks that humans could solve. It measures one capability family in March, not distance to AGI in October.
- Where
- Announcements
- Publisher
- ARC Prize Foundation
- Kind of source
- Benchmark result
- Describes events of
- Public from
Context only: ARC Prize Foundation,
AI agents struggled
Why it matters here: The preview found weaknesses in interactive exploration and planning. Its small game set and preliminary human baselines limit any inference about general intelligence or its proximity.
- Where
- 30-Day takeaways
- Publisher
- ARC Prize Foundation
- Kind of source
- Benchmark result
- Describes events of
- Public from
Context only: arXiv,
results on the data analysis exams
Why it matters here: The study reports gold-medal-level astronomy theory performance by Gemini 2.5 Pro and GPT-5, alongside uneven data-analysis performance and reasoning weaknesses. These strengths are material counterevidence to a uniformly weak-capability account, but do not establish AGI proximity.
- Where
- Abstract
- Publisher
- arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: Google DeepMind; ICML 2024,
future benchmarks
Why it matters here: This framework separates performance, generality, and autonomy. It supplies conceptual context but no threshold establishing whether October 2025 systems were close to AGI.
- Where
- Abstract
- Publisher
- Google DeepMind; ICML 2024
- Kind of source
- Peer-reviewed paper
- Describes events of
- Public from
What would settle it
A defensible operational meaning of close, paired with comprehensive evaluations of systems that existed by October 12, 2025. Later capabilities alone would not settle this historical observation.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, The Grand AGI Delusion
We aren’t close to AGI, but we are reshaping our world on the premise that we are.
- Where
- paragraph 1
- Published
- 2025-10-12
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-the-grand-agi-delusion-2aaae2-a1: Current. Outcome: Unresolved. 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.