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

52 claims

52 claims, sorted by newest said. Choose a claim to read it in full.
VerdictKindSaidClaimTopic
Mostly true Factual 25 Sep 2026 OpenAI's latest account called attacks "interactions" and only partially disclosed what happened. Factual · 25 Sep 2026 AI safety and alignment
Mostly true Factual 25 Sep 2026 OpenAI's software attacked a German web server. Factual · 25 Sep 2026 AI safety and alignment
Mostly true Factual 2 Jun 2026 More reports have recently come out questioning the return on investment of AI for corporate customers. Factual · 2 Jun 2026 Economics and markets
Mostly true Factual Jan 2026 Reviews by Joanna Stern and Marques Brownlee of one early humanoid robot prototype were damning. Factual · Jan 2026 AI capabilities
Mostly true Factual Jan 2026 Many would agree the AI field has reached a point of diminishing returns from scaling, and faith in scaling as a route to AGI has dissipated. Factual · Jan 2026 AI capabilities
Mostly true Factual Jan 2026 AGI did not materialize in 2025, contrary to predictions by Elon Musk and others, and GPT-5 was underwhelming and did not solve hallucinations. Factual · Jan 2026 AI timelines and forecasting
Mostly true Factual Jan 2026 AI agents did not turn out to be reliable in 2025. Factual · Jan 2026 AI capabilities
Mostly true Prediction 27 Jul 2025 Some of GPT-5's advice will be very bad. Prediction · 27 Jul 2025 AI capabilities
Mostly true Prediction 27 Jul 2025 GPT-5 will be considerably better than GPT-4. Prediction · 27 Jul 2025 AI capabilities
Mostly true Factual 1 Jan 2025 OpenAI has returned no profits, does not expect profits until at least 2029, has a burn rate in the billions, and lacks a clear technical moat. Factual · 1 Jan 2025 Economics and markets
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
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
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 Factual 2025 People put instructions in LLM system prompts but cannot get the systems to follow them. Factual · 2025 AI safety and alignment
Mostly true Factual 2025 LLMs, which are not AGI, are already eliminating some human work. Factual · 2025 AI, the economy and work
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 Factual 31 Mar 2024 In early August 2023 Gary Marcus wrote a piece titled 'What if Generative AI turned out to be a Dud', which was picked up by the Financial Times and other outlets. Factual · 31 Mar 2024 Their own work or organisation
Mostly true Factual 31 Mar 2024 Current-sized LLMs are expensive and not reliable enough for many jobs. Factual · 31 Mar 2024 Economics and markets
Mostly true Factual 31 Mar 2024 Hallucinations are inherent to how LLMs make predictions. Factual · 31 Mar 2024 AI capabilities
Mostly true Factual 31 Mar 2024 Marcus cited a statistic from a Wall Street Journal piece, "$50B in, $3B out", and said it was obviously not sustainable. Factual · 31 Mar 2024 Economics and markets
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
Mostly true Factual 26 Oct 2023 Some people, such as Marc Andreessen, are denying that there is any AI risk, long term or short term, and trying to turn whether to pursue AI safety into a debate. Factual · 26 Oct 2023 AI policy and governance
Mostly true Factual 26 Oct 2023 A newly published technique shows that current LLMs can be jailbroken quickly and automatically. Factual · 26 Oct 2023 AI safety and alignment
Mostly true Factual 26 Oct 2023 Gary Marcus emphasized a CERN for AI and global AI governance in April 2023 in his TED talk and in The Economist. Factual · 26 Oct 2023 Their own work or organisation
Mostly true Factual 30 Mar 2023 As of March 2023, none of the top AI labs is remotely transparent, AI governance basically doesn't exist, and there are basically no safeguards in place. Factual · 30 Mar 2023 AI policy and governance
Mostly true Factual 24 Mar 2023 The OpenAI and Microsoft GPT-4 papers reveal nothing about the model's architecture or training set. Factual · 24 Mar 2023 AI capabilities
Mostly true Factual 24 Mar 2023 Citing Ernie Davis, Marcus said OpenAI had begun to incorporate user experiments into its models' training corpus. Factual · 24 Mar 2023 AI capabilities
Mostly true Factual 24 Mar 2023 Gary Marcus said that on 2023-03-23 Microsoft had put out what he called a press release masquerading as science, which claimed that GPT-4 was an early (yet still incomplete) version of an AGI system. Factual · 24 Mar 2023 AI capabilities
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
Mostly true Factual 25 Dec 2022 ChatGPT regularly makes things up, and a fair bit of what it says is untrue. Factual · 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 Factual 25 Dec 2022 ChatGPT might sound like the Star Trek computer, but for now it still cannot be trusted. Factual · 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 Factual 25 Dec 2022 ChatGPT can write about anything but is easily confused. Factual · 25 Dec 2022 AI capabilities
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
Mostly true Factual Dec 2022 AI-generated misinformation became increasingly widespread, as Marcus warned in December 2022, but its effects have so far been less than he feared. Factual · Dec 2022 AI safety and alignment
Mostly true Factual Mar 2022 Gary Marcus's March 2022 article argued pure scaling would not solve hallucinations or abstraction and concluded there are serious holes in the scaling argument. Factual · Mar 2022 Their own work or organisation
Mostly true Factual 15 Nov 2021 Deep neural nets, nearly 10 years after their celebrated rebirth, are still utterly fooled by images like the one shown. Factual · 15 Nov 2021 AI capabilities
Mostly true Factual 12 May 2021 Biological brains have a high degree of precise internal structure that artificial neural networks lack. Factual · 12 May 2021 AI capabilities
Mostly true Factual 8 May 2021 Covid-19 is transmitted through the air (airborne). Factual · 8 May 2021 Biosecurity
Mostly true Factual 28 Oct 2020 As of 2020 there is no off-the-shelf software platform for building a domestic robot comparable to a game engine for building video games. Factual · 28 Oct 2020 Technology
Mostly true Factual 20 Sep 2019 Biological cognition involves many different mechanisms, working on different principles, that reinforce and counter-balance each other. Factual · 20 Sep 2019 Science and research methods
Mostly true Factual 20 Sep 2019 Deep learning does more than linear regression but is a more sophisticated form of statistical correlation. Factual · 20 Sep 2019 AI capabilities
Mostly true Factual 20 Sep 2019 Bad AI is leading to fatalities and undermining democracy. Factual · 20 Sep 2019 AI policy and governance
Back to the list

Gary Marcus

Open its page

Factual AI capabilities

“Despite all the hype, agents didn’t turn out to be reliable.”

Gary Marcus, author January 2026 Six (or seven) predictions for AI 2026 from a Generative AI realist (Marcus on AI)

AI agents did not turn out to be reliable in 2025.

Mostly true True in substance, wrong in a detail.

The second judge said resolved true; the verdict shown is the first judge's.

What would count as true or false

Judge as of January 2026, covering AI agents during 2025. True: 2025 evaluations and reports (agent benchmarks such as task-completion rates, industry surveys, documented deployment failures) show that general-purpose AI agents commonly failed at multi-step tasks and were not dependable enough to run unsupervised for broad use. Mostly true: agents were broadly unreliable, but some narrow categories (e.g., coding agents) achieved fairly dependable performance. Mixed: the evidence is evenly divided, with substantial reliable deployments alongside widespread failures. Mostly false: agents were largely reliable in mainstream use, and the remaining unreliability was limited. False: the evidence shows agents in 2025 were broadly reliable.

Evidence

  1. Year in review: The highs (and lows) of AI in 2025 The Indian Express · 31 December 2025
    “AI agents, widely seen as a step toward making AI companies more commercially viable, struggled to live up to expectations in 2025. While they might have shown some promise in tightly controlled enterprise settings, most AI agent-focused tools rolled out this year proved unreliable and inefficient for everyday tasks.”
  2. Why AI agents failed to take over in 2025 - it's 'a story as old as time,' says Deloitte - ZDNET ZDNET · 10 December 2025
    “Deloitte’s 2025 Emerging Technology Trends study, which surveys 500 US Tech Leaders, found that 30% of the surveyed organizations are exploring agentic options, with 38% piloting solutions and only 14% having solutions ready to deploy. The number of organizations actively using the systems in production is even lower, at 11%.”
  3. Gartner Survey Finds Just 15% of IT Application Leaders Are ... Gartner, Inc. · 30 September 2025
    “Seventy-five percent of survey respondents said they were piloting, were deploying or had already deployed some form of AI agents into their organization; however, concerns around governance, maturity and agent sprawl continue to hamper the deployment of truly agentic AI.”
  4. Developers remain willing but reluctant to use AI: The 2025 Developer Survey results are here Stack Overflow · 29 December 2025
    “We asked if the AI agent revolution was here, and the answer is a definitive "not yet." While 52% of developers say agents have affected how they complete their work, the primary benefit is personal productivity: 69% agree they've seen an increase.”

Reasoning

A year-end review [E1] states that most AI agent tools released in 2025 proved unreliable and inefficient for everyday tasks, while showing some promise in tightly controlled enterprise settings. Survey evidence fits limited maturity: Deloitte found only 11% of organisations using agents in production [E2]; Gartner reports that concerns about maturity and governance were holding back truly agentic deployment [E3]; and Stack Overflow's developer survey answered "not yet" to whether the agent revolution had arrived, while reporting that 52% of developers said agents had affected their work and 69% agreed they had seen a productivity increase [E4]. No task-completion benchmarks were provided. The evidence points to broad unreliability, with narrow pockets of usefulness, which fits mostly true.

Cite this claim

Gary Marcus, “AI agents did not turn out to be reliable in 2025.” (said January 2026). Verdict: mostly true. Reliability API, claim c_ffe59f328691, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_ffe59f328691

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.