“Any one of them would mean we need a real paradigm shift, if we are to get to AGI.”
Our reading, the sentence that was checked:
If any one of the three failure modes (not enough data, not enough compute, tasks that don't scale) holds, reaching AGI will need a real paradigm shift.
Assessment: Unresolved No deadline stated on 2026-10-29 Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, What “game over” for the latest paradigm in AI might look like, paragraph 16.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-what-game-over-for-the-latest-paradigm-b7b501-a2
We recorded 13 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 2022-10-29. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Conditional forecast
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No deadline stated
- Deadline state on 2026-10-29
- No deadline stated (the release’s reference date)
Conditions
- if any of the three failure modes (data, compute, non-scaling tasks) holds (antecedent, stated by the speaker)
- if we are to get to AGI (antecedent, stated by the speaker)
From the speaker
- Any one of them
- would mean we need a real paradigm shift
- if we are to get to AGI
Our interpretation
- 'Them' refers to insufficient data, insufficient available compute, or important tasks that do not improve sufficiently through scaling.
- A failure mode must be established as a binding obstacle to AGI through scaling. A temporary shortage, a projected bottleneck or a flat curve over a limited experiment does not by itself satisfy the antecedent.
- 'Real paradigm shift' means a fundamental change in approach beyond the additional scale and efficiencies allowed by the post's description of scaling maximalism.
- 'If we are to get to AGI' makes this a claim about what would be necessary for success, not a separate promise that AGI will arrive.
- AGI is read as broadly human-level general cognitive capability. No deadline is supplied.
Counts as supported if: At least one specified failure mode is established, and evidence demonstrates that overcoming it to reach AGI requires a fundamental change of approach.
Counts as contradicted if: At least one specified failure mode is established, yet independently validated AGI is achieved without a fundamental change of approach, through a route that actually encounters or overcomes that obstacle.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Did the forecast come true?
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 deadline stated
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
Epoch's June 6, 2024 analysis projects full use of a specified stock of human-generated public text between 2026 and 2032 if trends continue; it does not establish a binding data barrier to AGI [e1]. Its August 20, 2024 analysis projects that 2e29 FLOP training runs will likely be feasible by 2030, without establishing AGI's compute requirements [e2]. The semantics study revised July 17, 2024 reports imperfect entailment decoding above chance and revisits assumptions in the earlier theory [e3]. These items establish neither a specified binding obstacle nor a necessity proof or qualifying AGI demonstration, leaving the conditional claim unresolved [e1, e2, e3].
Limitations
- The data and compute projections do not establish whether or when either resource becomes a binding obstacle to AGI [e1, e2].
- The entailment results do not establish the limits of scaling across all important tasks or demonstrate AGI [e3].
- The packet does not show that a fundamental change of approach is necessary; successful use of an innovation alone would not establish necessity.
Evidence (3 items)
Context only: Epoch AI,
between 2026 and 2032
Why it matters here: This projects exhaustion of a specified stock of human-generated public text. It does not establish that the world has insufficient usable data for AGI or that such a constraint had occurred by the evaluation date.
- Where
- Opening summary
- Publisher
- Epoch AI
- Kind of source
- Independent measurement
- Describes events of
- Public from
Context only: Epoch AI,
will likely be feasible by 2030.
Why it matters here: The analysis identifies substantial constraints but projects room for further scaling. It neither proves that available compute is insufficient for AGI nor specifies how much compute AGI requires.
- Where
- Opening summary
- Publisher
- Epoch AI
- Kind of source
- Independent measurement
- Describes events of
- Public from
Context only: arXiv; William Merrill and colleagues,
implicitly model aspects of semantics
Why it matters here: The follow-up reports successful, imperfect entailment decoding and revisits assumptions in the earlier theory. It cautions against treating the earlier work as an established general data barrier, while not demonstrating AGI.
- Where
- Abstract
- Publisher
- arXiv; William Merrill and colleagues
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
Establish one specified failure mode as a binding obstacle to AGI through scaling. Then provide either a validated necessity result showing that reaching AGI requires a fundamental change of approach, or independently validated AGI achieved without such a change through a route that encounters or overcomes that obstacle.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, What “game over” for the latest paradigm in AI might look like
Any one of them would mean we need a real paradigm shift, if we are to get to AGI.
- Where
- paragraph 16
- Published
- 2022-10-29
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-what-game-over-for-the-latest-paradigm-b7b501-a2: 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.
About assessment marcus-what-game-over-for-the-latest-paradigm-b7b501-a2.