“Indeed, although it did not get the press it deserved, recent AAAI survey showed that the vast majority of academic AI researchers doubted that scaling would get us all the way to AGI. They were right.”
Our reading, the sentence that was checked:
A recent AAAI survey showed that the vast majority of academic AI researchers doubted that scaling would get us all the way to AGI, and they were right.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, OpenAI’s Waterloo? [with corrections], paragraph 22.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-openais-waterloo-fcefd7-a1
We recorded 9 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 2025-08-14. 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
- the vast majority of academic AI researchers (scope, stated by the speaker)
- all the way to AGI (qualifier, stated by the speaker)
From the speaker
- recent AAAI survey
- the vast majority of academic AI researchers
- doubted that scaling would get us all the way to AGI
- They were right.
Our interpretation
- The claim combines a report of researchers' opinions with an assertion that those opinions were correct. Both parts must be assessed.
- The ordinary reading of 'vast majority' is a clear supermajority. Responses of unlikely or very unlikely count as doubt; no numerical cutoff is imposed.
- Scaling means scaling the AI approaches discussed in the survey. The survey's scope must not be silently narrowed to LLM pretraining alone.
- The historical target is the recent survey and the claimed vindication by the August 14, 2025 post. The start is unspecified because 'recent' supplies no definite lower boundary.
- The plain reading of 'They were right' endorses the doubts about scaling's sufficiency for AGI. It is stronger than saying that one released model was not AGI.
- Survey agreement and the absence of AGI at a particular date cannot by themselves establish that scaling is insufficient.
Counts as supported if: The survey must report a clear supermajority doubting that scaling current approaches would achieve AGI, with evidence supporting the characterization of academic researchers. The additional endorsement requires evidence establishing the substantive limitation of scaling, beyond respondents' opinions or one model's shortcomings.
Counts as contradicted if: The survey must materially disagree with the reported finding or show that the claimed academic majority did not hold. Alternatively, a substantiated demonstration of AGI achieved through scaling the relevant approaches would refute the substantive endorsement. Missing subgroup data alone is not a contradiction.
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 survey component is substantially accurate: 76% of respondents rated scaling existing approaches to AGI unlikely or very unlikely. But this is an aggregate result from a predominantly academic sample, not a reported academic-only result. More importantly, the survey cannot establish 'They were right.' Later independent benchmark findings document substantial progress while expressly declining to establish AGI. The combined claim therefore remains unresolved. This is not mixed: support for one component and uncertainty about another do not constitute demonstrated contradiction.
Limitations
- The numerical quotation uses contemporaneous secondary reporting, checked against the original report, because read_page interleaves the primary PDF's columns.
- The survey's fieldwork dates and academic-only responses were not established.
- The March 7 availability date for e2 applies to the same information in the published report; the webpage itself is undated.
- Later benchmark evidence is context, outside the historical target period.
- Reading 'They were right' merely as reasonable skepticism would weaken the claim and could change its assessment.
Evidence (4 items)
Supports the claim: THE DECODER,
76 percent of respondents
Why it matters here: The preceding sentence identifies the possibility as scaling existing AI systems to AGI. This supports the reported supermajority among survey respondents, but does not establish an academic-only percentage or the correctness of their doubts.
- Publisher
- THE DECODER
- Kind of source
- News report
- Describes events of
- Public from
Context only: Association for the Advancement of Artificial Intelligence,
main affiliation (67%)
Why it matters here: The survey population was predominantly academic but included other affiliations. This sentence also appears in the March report available by March 7; that date bounds availability of the information, rather than dating the undated webpage.
- Publisher
- Association for the Advancement of Artificial Intelligence
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: AIhub,
has published
Why it matters here: This establishes that the report was publicly available before Marcus's post. It supplies a conservative availability bound for the report's introductory population information.
- Publisher
- AIhub
- Kind of source
- News report
- Describes events of
- Public from
Context only: ARC Prize,
not claiming
Why it matters here: ARC Prize reports substantial progress, including human-level action efficiency on its benchmark, but explicitly declines to identify the system as AGI. This later evidence challenges premature claims of vindication without decisively settling whether scaling is sufficient.
- Publisher
- ARC Prize
- Kind of source
- Independent measurement
- Describes events of
- Public from
What would settle it
Academic-only survey results and sampling information would settle the population qualification. The substantive endorsement needs a clear AGI criterion and decisive evidence about scaling's sufficiency: for example, a validated AGI demonstration with documented methods, or a technical limitation shown to apply to the relevant approaches. Continued absence of AGI alone would not settle it.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, OpenAI’s Waterloo? [with corrections]
Indeed, although it did not get the press it deserved, recent AAAI survey showed that the vast majority of academic AI researchers doubted that scaling would get us all the way to AGI. They were right.
- Where
- paragraph 22
- Published
- 2025-08-14
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-openais-waterloo-fcefd7-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.