“Even if the AI acceleration we saw from 2019-2023 slows down for five years, as I have often speculated, it won’t slow down forever.”
Our reading, the sentence that was checked:
The AI acceleration seen from 2019-2023 will not slow down forever, even if it slows down for five years as Marcus has often speculated.
Assessment: Unresolved No deadline stated on 2026-10-29 Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Hard-forked! Casey Newton’s distorted portrait of Gary Marcus and AI skepticism, paragraph 8.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-hard-forked-casey-newtons-distorted-637742-a2
We recorded 14 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 2024-12-06. 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)
Conditions
- even if the 2019-2023 acceleration slows down for five years, as he has often speculated (qualifier, stated by the speaker)
From the speaker
- Even if
- the AI acceleration we saw from 2019-2023
- slows down for five years
- it won’t slow down forever
Our interpretation
- Assess the conditional proposition that a slowdown, if it occurs, eventually ends.
- Even if does not assert that a slowdown will occur. Five years describes a hypothetical duration, not a deadline for renewed acceleration.
- Renewed acceleration means a sustained increase in the rate of capability progress after an established slowdown, measured on comparable tasks.
- The alternative concessive reading makes the eventual-resumption statement broader; it still supplies no deadline.
Counts as supported if: After a documented slowdown relative to the 2019–2023 pace, comparable longitudinal evidence establishes renewed acceleration.
Counts as contradicted if: A decisive event establishes that progress can never accelerate again. A finite period of continued slowdown cannot establish forever.
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
METR's July 14, 2025 report found task time horizons doubling roughly every seven months, with possible acceleration to every four months in 2024 [e3]. This does not establish a slowdown relative to 2019–2023 followed by sustained renewed acceleration [e3]. A March 20, 2026 research note also reports increasing sensitivity to analysis choices as the task suite saturates [e4]. These measurements do not settle the eventual-resumption claim [e3, e4].
Limitations
- No consistent series establishes the required baseline, slowdown and subsequent sustained acceleration.
- Task time horizons do not measure all AI progress, and estimates depend on task selection and analysis choices.
- The hypothetical five-year slowdown supplies no deadline; a finite slowdown cannot establish that slowdown lasts forever.
Evidence (2 items)
Context only: METR,
possibly accelerating
Why it matters here: The retrospective measurements suggest faster progress on this measure in 2024. They do not establish a slowdown followed by recovery, or a five-year slowdown.
- Publisher
- METR
- Kind of source
- Independent measurement
- Describes events of
- Public from
Context only: METR; Alexander Barry,
more sensitive to analysis choices
Why it matters here: This qualification limits how confidently changes in measured task horizons can establish a broad change in the pace of AI progress.
- Publisher
- METR; Alexander Barry
- Kind of source
- Independent measurement
- Describes events of
- Public from
What would settle it
Independent longitudinal measurements with stable methods and uncertainty estimates establishing a slowdown relative to the 2019–2023 baseline, followed by sustained renewed acceleration on comparable tasks; or a decisive event establishing that future acceleration is impossible.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Hard-forked! Casey Newton’s distorted portrait of Gary Marcus and AI skepticism
Even if the AI acceleration we saw from 2019-2023 slows down for five years, as I have often speculated, it won’t slow down forever.
- Where
- paragraph 8
- Published
- 2024-12-06
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-hard-forked-casey-newtons-distorted-637742-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-hard-forked-casey-newtons-distorted-637742-a2.