ClaimsPeople

Gary Marcus

Gary Marcus is an author and AI entrepreneur and professor emeritus at NYU; he has written six books, including Rebooting AI and Taming Silicon Valley, founded Geometric Intelligence (bought by Uber), testified at the May 2023 US Senate AI hearing, and writes a newsletter with over 121,000 subscribers.

712 claims collected, 213 researched, 103 resolved. Report of 4 November 2026, method 0.1.0+c876030c.

Scores by kind of claim

Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.

  • Factual claims

    Statements about the present or the past, checked against the record: did they hold up?

    79 72 85

    High confidence, 64 factual claims resolved. From 18 documents, 2019 to 2026; 19 documents sampled from 37 candidates found.

    Not scored: 32 unresolvable, 4 checkable only against their own records.

  • Predictions

    Statements about events outside the person's control, checked once they fell due: did they come true?

    80 72 88

    Medium confidence, 38 predictions resolved. From 7 documents, 2019 to 2025; 19 documents sampled from 37 candidates found.

    Not scored: 60 unresolvable. 13 more fall due later.

    Does not yet allow for how likely each prediction was when made.

  • Promises

    Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?

    Insufficient record: 1 promise resolved, 5 needed for a score.

    Not scored: 1 unresolvable.

All types combined: 79 (likely 74 to 84, high confidence), from 102 claims: 64 factual claims, 38 predictions. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.

Gary Marcus's claims

38 claims, sorted by newest said. Choose a claim to read it in full.
VerdictKindSaidClaimTopic
Resolved true Prediction 27 Jul 2025 OpenAI will release GPT-5 imminently. Prediction · 27 Jul 2025 AI capabilities
Resolved true Prediction 27 Jul 2025 GPT-5 will still make a significant number of obvious errors that are hard to predict. Prediction · 27 Jul 2025 AI capabilities
Resolved true Prediction 27 Jul 2025 If the model being discussed has guardrails like ChatGPT's, they will teeter between being too weak (beaten by jailbreaks) and too strong (rejecting some perfectly reasonable requests). Prediction · 27 Jul 2025 AI safety and alignment
Mostly true Prediction 27 Jul 2025 Some of GPT-5's advice will be very bad. Prediction · 27 Jul 2025 AI capabilities
Resolved true Prediction 27 Jul 2025 GPT-5 by itself will not be a general-purpose AGI capable of arbitrary tasks. Prediction · 27 Jul 2025 AI timelines and forecasting
Mostly true Prediction 27 Jul 2025 GPT-5 will be considerably better than GPT-4. Prediction · 27 Jul 2025 AI capabilities
Resolved true Prediction 1 Jan 2025 AI Safety Institutes will offer guidance in 2025 but have little legal authority to stop truly dangerous models. Prediction · 1 Jan 2025 AI policy and governance
Resolved true Prediction 1 Jan 2025 Profits from AI models will remain modest or nonexistent in 2025. Prediction · 1 Jan 2025 Economics and markets
Resolved true Prediction 1 Jan 2025 Lack of reliability will continue to afflict generative AI in 2025. Prediction · 1 Jan 2025 AI capabilities
Resolved true Prediction 1 Jan 2025 Artificial general intelligence will not be achieved in 2025. Prediction · 1 Jan 2025 AI timelines and forecasting
Resolved true Prediction 1 Jan 2025 Hallucinations will continue to afflict generative AI in 2025. Prediction · 1 Jan 2025 AI capabilities
Resolved true Prediction 1 Jan 2025 Less than 10% of the workforce will be replaced by AI in 2025, probably less than 5%. Prediction · 1 Jan 2025 AI, the economy and work
Resolved true Prediction 1 Jan 2025 The US will continue to have very little regulation protecting consumers from generative AI risks in 2025. Prediction · 1 Jan 2025 AI policy and governance
Mostly true Prediction 1 Jan 2025 A large-scale cyberattack in which generative AI plays an important causal role may well occur in 2025. Prediction · 1 Jan 2025 Other global risks
Resolved true Prediction 1 Jan 2025 AI agents will be heavily hyped throughout 2025 but remain far from reliable, except possibly in very narrow use cases. Prediction · 1 Jan 2025 AI capabilities
Resolved true Prediction 1 Jan 2025 AI power consumption will rise in 2025 and remain a major problem, while few generative AI companies will be transparent about their usage. Prediction · 1 Jan 2025 AI compute, hardware and data
Mostly true Prediction 1 Jan 2025 Few if any radiologists will be replaced by AI in 2025. Prediction · 1 Jan 2025 AI, the economy and work
Resolved true Prediction 1 Jan 2025 No single AI system will solve more than 4 of the AI 2027 Marcus-Brundage tasks by the end of 2025. Prediction · 1 Jan 2025 AI capabilities
Resolved true Prediction 1 Jan 2025 In 2025 humanoid robotics will see a lot of hype, but nobody will release a robot remotely as capable as Rosie the Robot. Prediction · 1 Jan 2025 AI capabilities
Mostly true Prediction 1 Jan 2025 Neurosymbolic AI will become much more prominent in 2025. Prediction · 1 Jan 2025 AI capabilities
Mostly true Prediction 1 Jan 2025 OpenAI's Sora will continue to have trouble with physics in 2025. Prediction · 1 Jan 2025 AI capabilities
Mostly true Prediction 31 Mar 2024 Marcus doubted that a mindblowing GPT model would be released in 2024. Prediction · 31 Mar 2024 AI timelines and forecasting
Mostly true Prediction 31 Mar 2024 GPT-5 is likely to be even more expensive to train and operate than GPT-4. Prediction · 31 Mar 2024 Economics and markets
Mixed Prediction 26 Oct 2023 With more than 60 elections around the world in 2024, the chance that at least one would be swayed by deepfakes is high. Prediction · 26 Oct 2023 Other global risks
Mostly true Prediction 25 Dec 2022 GPT-4 will still be reckless and hard to control, making a significant number of stupid errors. Prediction · 25 Dec 2022 AI capabilities
Mostly true Prediction 25 Dec 2022 GPT-4's natural language output still won't be reliably usable as input to downstream programs such as databases or virtual assistants. Prediction · 25 Dec 2022 AI capabilities
Resolved true Prediction 25 Dec 2022 GPT-4 will definitely seem smarter than its predecessors, but its internal architecture will remain problematic. Prediction · 25 Dec 2022 AI capabilities
Mostly true Prediction 25 Dec 2022 GPT-4 will not be trustworthy and complete enough to give reliable medical advice. Prediction · 25 Dec 2022 AI capabilities
Mostly true Prediction 25 Dec 2022 GPT-4 will be a very large, powerful system (a 'monster'). Prediction · 25 Dec 2022 AI capabilities
Mostly true Prediction 25 Dec 2022 With GPT-4, fluent hallucinations will still be common and easily induced, escalating the risk of LLMs being used to create misinformation. Prediction · 25 Dec 2022 AI capabilities
Mostly true Prediction 25 Dec 2022 GPT-4 may be better at faking term papers, but if it follows the same playbook as its predecessors, it still won't really understand the world, and its flaws will eventually show. Prediction · 25 Dec 2022 AI capabilities
Mostly true Prediction 25 Dec 2022 GPT-4 will blow people's minds when released. Prediction · 25 Dec 2022 AI capabilities
Mostly true Prediction 25 Dec 2022 Marcus suspected that GPT-4's release would follow a pattern of immense buzz, then scientific scrutiny, then recognition that many problems remain. Prediction · 25 Dec 2022 AI capabilities
Mostly true Prediction 25 Dec 2022 If GPT-4 is architecturally essentially the same as GPT-3, it can be expected to be unable to construct internal models of how the world works and, in consequence, unable to understand things at an abstract level. Prediction · 25 Dec 2022 AI capabilities
Mostly true Prediction 25 Dec 2022 AI alignment between human wants and machine behavior will remain a critical unsolved problem, with GPT-4 unable to reliably follow shared human values. Prediction · 25 Dec 2022 AI safety and alignment
Mostly true Prediction 25 Dec 2022 GPT-4 will remain a pastiche generator and useful for brainstorming and first drafts, but not trustworthy general intelligence. Prediction · 25 Dec 2022 AI capabilities
Resolved false Prediction 14 Dec 2021 Gary Marcus bet that hybrid (neurosymbolic) AI models would pull ahead in 2022. Prediction · 14 Dec 2021 AI timelines and forecasting
Resolved true Prediction 20 Sep 2019 Marcus said he did not think machines would reach world-class scientific reasoning within the next five years (by 2024). Prediction · 20 Sep 2019 AI timelines and forecasting
Back to the list

Gary Marcus

Open its page

Prediction AI safety and alignment

““Alignment” between what humans want and what machines do will continue to be a critical, unsolved problem.”

Gary Marcus, author 25 December 2022 What to Expect When You're Expecting … GPT-4 (Marcus on AI)

AI alignment between human wants and machine behavior will remain a critical unsolved problem, with GPT-4 unable to reliably follow shared human values.

Mostly true True in substance, wrong in a detail.

What would count as true or false

Judge against GPT-4 as released (March 2023) and the state of the field during 2023. True: through 2023 alignment remained widely recognised as an unsolved, critical problem, and GPT-4 demonstrably failed to reliably follow human values and intentions, e.g. jailbreaks, harmful outputs, documented misbehaviour in OpenAI's system card or independent red-teaming. Mostly true: both parts hold, but GPT-4's failures were notably less frequent than the claim implies (e.g. it mostly behaved well, with failures mainly under adversarial prompting). Mixed: one part holds and the other does not, e.g. alignment remained unsolved in general but GPT-4 was shown to follow shared human values reliably in practice. Mostly false: GPT-4 was broadly reliable in following human values with only rare, narrow failures, and experts considered practical alignment largely achieved for such models. False: alignment was widely regarded as solved and GPT-4 reliably followed human values.

Evidence

  1. GPT-4 System Card - cdn.openai.com OpenAI
    “Training on this data improves model robustness but does not fully solve the problem of “jailbreaks” leading to harmful content.”
  2. GPT-4 System Card - cdn.openai.com OpenAI
    “In most cases, GPT-4-launch exhibits much safer behavior due to the safety mitigations we applied.”
  3. Model alignment protects against accidental harms, not intentional ones normaltech.ai (Arvind Narayanan, Sayash Kapoor, Seth Lazar) · 1 December 2023
    “RLHF has very substantially solved this problem, and by solving it OpenAI created a multi-billion dollar industry where one didn't exist before.”
  4. Model alignment protects against accidental harms, not intentional ones normaltech.ai (Arvind Narayanan, Sayash Kapoor, Seth Lazar) · 1 December 2023
    “A wave of recent research has shown that RLHF is brittle.”

Reasoning

OpenAI's GPT-4 System Card says its safety training "does not fully solve the problem of 'jailbreaks' leading to harmful content" (E1), and later independent analysis calls RLHF "brittle" (E4). Both support the claim that alignment stayed an unsolved problem. However, the same system card says the launch version of GPT-4 showed "much safer behavior" in most cases (E2), and Narayanan, Kapoor and Lazar write that RLHF "very substantially solved" the problem of accidental harms (E3). The failures therefore appear mainly under adversarial prompting rather than in ordinary use, which matches the criteria's description of "mostly true".

Cite this claim

Gary Marcus, “AI alignment between human wants and machine behavior will remain a critical unsolved problem, with GPT-4 unable to reliably follow shared human values.” (said 25 December 2022). Verdict: mostly true. Reliability API, claim c_d30cb5d57cfa, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_d30cb5d57cfa

Claims without a verdict

Is a verdict here wrong, or do you speak for Gary Marcus? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.