“I couldn’t agree more. And parallel arguments could be made in many domains.”
Our reading, the sentence that was checked:
Marcus endorses the view that ChatGPT and similar tools will not themselves be virtual scientists anytime soon, and says parallel arguments could be made in many domains.
Assessment: Unresolved No deadline stated on 2026-10-29 Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Superhuman AGI is not nigh, paragraph 29.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-superhuman-agi-is-not-nigh-a1e644-a2
We recorded 13 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2024-04-12. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Forecast
- Deadline, as the speaker wrote it
- “anytime soon”
- Deadline state when assessed
- No deadline stated
- Deadline state on 2026-10-29
- No deadline stated (the release’s reference date)
Conditions
- parallel arguments could be made in many domains (qualifier, stated by the speaker)
From the speaker
- I couldn’t agree more.
- ChatGPT and its relatives
- they will not themselves, anytime soon, be virtual scientists
- parallel arguments could be made in many domains
Our interpretation
- The endorsement adopts the quoted forecast about ChatGPT-like systems, including systems using language models with tools.
- The ordinary reading of virtual scientists is systems that carry out scientific investigations and produce valid new findings, beyond summarizing literature.
- A qualifying investigation may use a human-supplied broad objective, existing data, software tools, and subsequent human validation. It need not involve physical laboratory work or a Nobel-caliber discovery.
- Humans supplying the central hypothesis, directing the essential scientific reasoning, or repairing the finding would weaken a claim that the system did the investigation itself.
- Anytime soon provides no definite deadline. Do not substitute the end-of-2025 deadline attached to the surrounding task list.
- Parallel arguments could be made in many domains is an analogy, not a conditional antecedent. The unnamed domains do not supply a separately measurable forecast.
Counts as supported if: Evidence that systems remain scientific assistants would be consistent with the forecast, but cannot conclusively support an indefinite negative prediction without an agreed endpoint for anytime soon.
Counts as contradicted if: A dated, transparent demonstration shows a ChatGPT-like system carrying out an investigation that produces a novel finding validated by independent scientific assessment, with human contributions disclosed, sufficiently soon after April 12, 2024 to satisfy the temporal wording.
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
Sakana's March 12, 2025 report describes end-to-end AI research, human selection of submissions, and favorable workshop scores for one paper [e5]. The Kosmos preprint available November 5, 2025 reports four novel contributions and domain-expert corroboration within the developer-led study [e7]. The Nature account published March 25, 2026 documents automated research and a generated manuscript passing the first workshop-review round [e8]. These records show substantial research automation, but do not unambiguously establish the combination of autonomy, independently validated novelty, and timing required by the contract's early-resolution rule [e5, e7, e8]. The forecast remains unresolved.
Limitations
- “Anytime soon” supplies no deadline; reaching the evaluation date cannot establish support.
- Favorable manuscript reviews and peer review of a system report do not independently verify each generated finding.
- Kosmos's reported expert corroboration does not settle all questions about independent validation and human contributions.
- The evaluation of the earlier AI Scientist does not establish the limits of later versions [e6].
- The unnamed additional domains cannot be assessed separately.
Evidence (4 items)
Context only: Sakana AI,
without any modifications from humans.
Why it matters here: Sakana reports autonomous hypothesis generation, experiments, analysis, and writing, followed by human selection of three submissions and favorable workshop scores for one. This is substantial counterevidence, but the paper was withdrawn without final meta-review, and the account alone does not independently validate its scientific finding.
- Where
- Evaluation Process
- Publisher
- Sakana AI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: Joeran Beel, Min-Yen Kan, and Moritz Baumgart; arXiv,
hallucinated numerical results.
Why it matters here: An independent evaluation of the earlier AI Scientist found unreliable novelty assessments, failed experiments, and unsupported results. It limits what manuscript production alone establishes, but does not disprove capabilities of later versions; available_at identifies the revised version read.
- Publisher
- Joeran Beel, Min-Yen Kan, and Moritz Baumgart; arXiv
- Kind of source
- Independent measurement
- Describes events of
- Public from
Context only: Ludovico Mitchener and colleagues; arXiv,
independent analysis
Why it matters here: Kosmos's authors report autonomous analysis from supplied objectives and datasets, including new findings corroborated by domain experts. This is a serious candidate contradiction, but the corroboration is reported within the developer-led study, and whether late-2025 achievements satisfy anytime soon remains unspecified.
- Publisher
- Ludovico Mitchener and colleagues; arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Context only: Nature; Chris Lu and colleagues,
automatically test, report on and evaluate
Why it matters here: The peer-reviewed account documents automated research in template-based and template-free settings, and describes the generated paper's success as passing the first workshop-review round. It strengthens evidence for substantial research autonomy, while peer review of this system report does not independently reproduce each generated finding.
- Where
- Abstract
- Publisher
- Nature; Chris Lu and colleagues
- Kind of source
- Peer-reviewed paper
- Describes events of
- Public from
What would settle it
Transparent investigation records disclosing human contributions, together with independent scientific validation of a novel finding and unambiguous qualification under “anytime soon,” could establish contradiction. Otherwise, a defensible endpoint and operational definition of “virtual scientists” are needed to assess the forecast over a defined period.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Superhuman AGI is not nigh
I couldn’t agree more. And parallel arguments could be made in many domains.
- Where
- paragraph 29
- Published
- 2024-04-12
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-superhuman-agi-is-not-nigh-a1e644-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-superhuman-agi-is-not-nigh-a1e644-a2.