““Scaling” alone—building bigger and models until they absorb the entire internet — will prove useful, but only to a point.”
Our reading, the sentence that was checked:
Scaling alone will prove useful only up to a point.
Assessment: Unresolved No deadline stated on 2026-10-29 Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, What to Expect When You’re Expecting … GPT-5, paragraph 9.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-what-to-expect-when-youre-expecting-62e-ba0b73-a1
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 2025-07-27. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- 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)
From the speaker
- “Scaling” alone
- building bigger and models until they absorb the entire internet
- will prove useful
- but only to a point
Our interpretation
- Scaling means increasing language-model size, training data, and associated training compute without adding qualitatively different methods.
- In context, only to a point means insufficient by itself to deliver trustworthy general intelligence.
- The nearby decade forecast concerns a shift in research emphasis and does not supply this claim's deadline.
- A current plateau, an individual model failure, or adoption of additional techniques does not establish an ultimate scaling ceiling.
Counts as supported if: Evidence establishes useful gains from scaling and a binding limitation preventing scaling alone from reaching trustworthy general intelligence.
Counts as contradicted if: An independently validated system reaches trustworthy general intelligence through scaling alone, with architectural and training disclosures sufficient to establish that attribution.
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
Scaling studies document useful gains, and research on inconsistent errors gives reasons to doubt that scale alone solves reliability. Neither establishes the ultimate ceiling Marcus predicts. Existing failures and the use of hybrid systems cannot settle that question. With no deadline and no decisive event, the forecast remains open.
Limitations
- Finite experiments cannot establish what all future increases in scale will achieve.
- The endpoint and the meaning of scaling alone remain underspecified.
Evidence (2 items)
Context only: arXiv,
scaling laws
Why it matters here: The study relates model scale and training to measurable performance across a finite experimental range. It supports scaling's usefulness but does not establish its ultimate sufficiency or limit.
- Where
- Abstract
- Publisher
- arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: arXiv,
Consequently, scale alone seems unlikely to eliminate error-incoherence.
Why it matters here: The paper reports that the relationship between scale and inconsistent errors depends on the experiment. Its qualified inference concerns error behavior in tested settings, not a demonstrated upper bound on scaling toward trustworthy general intelligence.
- Where
- Abstract, version 2
- Publisher
- arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
An independently verified scaling-only system achieving a clearly defined standard of trustworthy general intelligence, or a validated binding limitation showing why scaling alone cannot reach that standard.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, What to Expect When You’re Expecting … GPT-5
“Scaling” alone—building bigger and models until they absorb the entire internet — will prove useful, but only to a point.
- Where
- paragraph 9
- Published
- 2025-07-27
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-what-to-expect-when-youre-expecting-62e-ba0b73-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.
About assessment marcus-what-to-expect-when-youre-expecting-62e-ba0b73-a1.