Pundit ClaimsAI claims by Gary Marcus and Ed Zitron, with the evidence

Changes, page 33 of 37

Every substantive change to the register, newest first. Speaker events record what the people covered said or changed. Site events record what this site published, corrected, flagged, or withdrew. The two never share a label.

Substantive changes (7383)

Page 33 of 37: changes 6401 to 6600 of 7383, newest first. Dates are when the change was recorded.

  1. Assessment publishedBy this siteGary Marcus

    OpenAI may well end up as the WeWork of AI.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-and-how-the-end-might-31d45a-a1.

  2. Assessment publishedBy this siteGary Marcus

    Banks evidently do not think OpenAI is worth its current valuation and don't want its stock as collateral.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-and-how-the-end-might-2d0a21-a1.

  3. Assessment publishedBy this siteGary Marcus

    Anthropic has taken a large share of the business market while OpenAI has lost much of it.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-and-how-the-end-might-2b3f1d-a2.

  4. HeldBy this siteGary Marcus

    Held record marcus-breaking-news-and-how-the-end-might-220a31

    e1 measures something else (metric Inference (model running) cost is not Cost to run (inference/serving cost) of latest models); it can only be context unless it is derived with stated inputs; e2 measures something else (metric Adjusted gross margin (pressured by inference cost) is not Cost to run (inference/serving cost) of latest models); it can only be context unless it is derived with stated inputs

  5. HeldBy this siteGary Marcus

    Held record marcus-breaking-news-and-how-the-end-might-1da588

    The final-text check failed; overclaims: The claim depends on a 'reliable version of the product', which Marcus does not define and which he generally holds does not yet exist. | Provider margins on current products can be observed but are not the product he describes, so they are context only..

  6. Assessment publishedBy this siteGary Marcus

    OpenAI has not yet turned a profit and it is unclear whether it ever will.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-and-how-the-end-might-13c877-a2.

  7. Assessment publishedBy this siteGary Marcus

    At some point OpenAI will be unable to meet its financial obligations.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-and-how-the-end-might-0d7327-a2.

  8. HeldBy this siteGary Marcus

    Held record marcus-breaking-news-and-how-the-end-might-0ac250

    The final-text check failed; overclaims: That does not fit a failure to meet its obligations..

  9. Assessment publishedBy this siteGary Marcus

    The most likely scenario for the AI boom breaking is that OpenAI at some point can't make ends meet.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-and-how-the-end-might-01f1b4-a1.

  10. Assessment publishedBy this siteGary Marcus

    If METR's finding that AI tools slowed down task completion is general and replicable, it is a serious blow to generative AI's flagship use case.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-ai-coding-may-not-be-b63764-a1.

  11. Assessment publishedBy this siteGary Marcus

    The gap between coders' expected and actual efficiency gains from AI tools warrants reevaluating the tools' value.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-ai-coding-may-not-be-af81ea-a1.

  12. Assessment publishedBy this siteGary Marcus

    When pointing to positive uses of generative AI, Marcus invariably points to how it serves as kind of a valuable autocomplete for coding.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-ai-coding-may-not-be-a03dc2-a1.

  13. Assessment publishedBy this siteGary Marcus

    People using AI coding tools may be imagining productivity gains they are not getting while ignoring real-world costs.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-ai-coding-may-not-be-87b039-a1.

  14. Assessment publishedBy this siteGary Marcus

    A significant fraction of the money people have paid for generative AI has been for coding assistance.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-ai-coding-may-not-be-3a9e63-a2.

  15. Assessment publishedBy this siteGary Marcus

    One possible mechanism behind the AI-coding slowdown is that debugging code you didn't write (AI-generated code) can be hard.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-ai-coding-may-not-be-25a28a-a1.

  16. HeldBy this siteGary Marcus

    Held record marcus-breaking-news-ai-coding-may-not-be-153c1b

    The anthropic screen reads the claim differently.

  17. HeldBy this siteGary Marcus

    Held record marcus-breaking-news-agi-is-not-imminent-bf2312

    The openai review does not accept e3 as supports.

  18. Assessment publishedBy this siteGary Marcus

    On 2025-01-20 Sam Altman walked back his earlier hint that OpenAI was close to building AGI.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-agi-is-not-imminent-9df7cc-a1.

  19. Assessment publishedBy this siteGary Marcus

    Marcus says Altman's statement that OpenAI is confident it knows how to build AGI gave him, to be honest, Elizabeth Holmes vibes.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-news-agi-is-not-imminent-1dfec7-a1.

  20. HeldBy this siteGary Marcus

    Held record marcus-breaking-marcus-weighs-in-mostly-fca2d9

    The openai screen doubts the attribution.

  21. HeldBy this siteGary Marcus

    Held record marcus-breaking-marcus-weighs-in-mostly-da880f

    The final-text check failed: proposition_faithful, conditions_shown; overclaims: AI systems need explicit, structured world knowledge to avoid hallucinations..

  22. Assessment publishedBy this siteGary Marcus

    LeCun's new company pursuing a new approach to AI will probably not succeed.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-marcus-weighs-in-mostly-d17548-a1.

  23. HeldBy this siteGary Marcus

    Held record marcus-breaking-marcus-weighs-in-mostly-c3d20e

    The anthropic screen reads the claim differently.

  24. HeldBy this siteGary Marcus

    Held record marcus-breaking-marcus-weighs-in-mostly-b4a926

    The openai review found the criteria unfair to the speaker's words.

  25. HeldBy this siteGary Marcus

    Held record marcus-breaking-marcus-weighs-in-mostly-751464

    The anthropic review found the criteria unfair to the speaker's words.

  27. HeldBy this siteGary Marcus

    Held record marcus-breaking-marcus-weighs-in-mostly-6c50c5

    The final-text check returned an unusable answer for this claim.

  28. HeldBy this siteGary Marcus

    Held record marcus-breaking-marcus-weighs-in-mostly-679433

    The anthropic review found the criteria unfair to the speaker's words.

  30. Assessment publishedBy this siteGary Marcus

    If LeCun keeps failing to credit others' work at his new company, morale there will be very low.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-marcus-weighs-in-mostly-5998ea-a2.

  31. Assessment publishedBy this siteGary Marcus

    LeCun's new company is trying to explore something genuinely new, and that is what the AI field desperately needs.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-marcus-weighs-in-mostly-4e34de-a1.

  32. HeldBy this siteGary Marcus

    Held record marcus-breaking-marcus-weighs-in-mostly-2b53b6

    The openai review found the criteria unfair to the speaker's words.

  33. Assessment publishedBy this siteGary Marcus

    LeCun's world-model approach will likely fall short.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-marcus-weighs-in-mostly-25749b-a1.

  34. HeldBy this siteGary Marcus

    Held record marcus-breaking-llm-reasoning-continues-edef38

    only a conditional forecast records a condition status

  35. HeldBy this siteGary Marcus

    Held record marcus-breaking-llm-reasoning-continues-de8c76

    The two reviews differ: anthropic historical_observation/not_applicable/supported, openai historical_observation/not_applicable/unresolved.

  36. Assessment publishedBy this siteGary Marcus

    The Caltech/Stanford review of LLM reasoning failures shows how far AI still has to go and, implicitly, why LLMs aren't the right answer.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-llm-reasoning-continues-8789a8-a1.

  37. Assessment publishedBy this siteGary Marcus

    The AI industry should face reality and start focusing on alternatives to LLMs instead of dismissing critics.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-llm-reasoning-continues-6ec345-a1.

  38. Assessment publishedBy this siteGary Marcus

    As of February 2026, reasoning is still one of the core unsolved problems for LLMs.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-llm-reasoning-continues-4f61c1-a2.

  39. Assessment publishedBy this siteGary Marcus

    As of February 2026, LLMs still hallucinate and still make boneheaded errors.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-llm-reasoning-continues-4f542d-a2.

  42. Assessment publishedBy this siteGary Marcus

    An injunction against OpenAI is needed immediately; without one, society risks a massive and possibly deadly AI-induced infrastructure problem.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-florida-seeks-injunction-f055ee-a1.

  43. Assessment publishedBy this siteGary Marcus

    Every state and country should follow Florida in seeking to pause OpenAI's ChatGPT development.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-florida-seeks-injunction-b9aff3-a1.

  44. Assessment publishedBy this siteGary Marcus

    LLMs on their own should never be trusted, so it is right to hand critical agent-safety functions to deterministic symbolic code.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-florida-seeks-injunction-b81139-a1.

  45. HeldBy this siteGary Marcus

    Held record marcus-breaking-florida-seeks-injunction-89dcd4

    The anthropic screen reads the claim differently.

  46. Assessment publishedBy this siteGary Marcus

    Nvidia often announces software, such as NemoClaw, that has had relatively little impact.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-florida-seeks-injunction-89bb07-a2.

  47. Assessment publishedBy this siteGary Marcus

    Two and a half months after NemoClaw was announced, probably nobody reading this post has even heard of it.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-florida-seeks-injunction-33d837-a1.

  48. Assessment publishedBy this siteGary Marcus

    OpenAI lacks the ability to properly regulate its own technology.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-florida-seeks-injunction-240d28-a2.

  49. HeldBy this siteGary Marcus

    Held record marcus-breaking-expensive-new-evidence-that-fc9314

    The anthropic review reads the claim differently.

  51. HeldBy this siteGary Marcus

    Held record marcus-breaking-expensive-new-evidence-that-a5e631

    The anthropic screen finds the claim checkable, so it must be graded.

  52. HeldBy this siteGary Marcus

    Held record marcus-breaking-expensive-new-evidence-that-7cde5b

    The openai screen reads the claim differently.

  54. HeldBy this siteGary Marcus

    Held record marcus-breaking-expensive-new-evidence-that-77675a

    a observation claim cannot be assessed on the causal_assessment axis

  55. HeldBy this siteGary Marcus

    Held record marcus-breaking-expensive-new-evidence-that-76f449

    The openai screen reads the claim differently.

  56. Assessment publishedBy this siteGary Marcus

    Marcus dismisses pure scaling in light of the Meta and xAI results ('so much for pure scaling').

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-expensive-new-evidence-that-68a1ab-a1.

  57. HeldBy this siteGary Marcus

    Held record marcus-breaking-expensive-new-evidence-that-3c46bc

    The openai screen reads the claim differently.

  58. HeldBy this siteGary Marcus

    Held record marcus-breaking-demis-hassabis-endorses-ead065

    The final-text check failed; overclaims: It also notes Hassabis hopes to launch the body by year's end, which supports 'strongly'..

  61. Assessment publishedBy this siteGary Marcus

    The FINRA model Hassabis suggests is excellent, and the world will be a better place if it is implemented with transparency and independence.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-demis-hassabis-endorses-8f3160-a1.

  63. HeldBy this siteGary Marcus

    Held record marcus-breaking-demis-hassabis-endorses-246dae

    The two reviews differ: anthropic historical_observation/not_applicable/supported, openai historical_observation/not_applicable/unresolved.

  65. Assessment publishedBy this siteGary Marcus

    Nvidia's new truly open-source Nemotron model could ultimately undercut OpenAI and Anthropic, companies Nvidia has deals with.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-circular-financing-reaches-37142e-a1.

  66. Assessment publishedBy this siteGary Marcus

    Nvidia's new truly open-source (not just open-weight) Nemotron release is good for end users and developers.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-circular-financing-reaches-37142e-2-a1.

  67. Assessment publishedBy this siteGary Marcus

    The change replacing AB-501's content with unrelated provisions is real and not a clerical error, according to the author's sources.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-bill-that-would-have-blocked-e557a4-a2.

  68. HeldBy this siteGary Marcus

    Held record marcus-breaking-bill-that-would-have-blocked-c89fdc

    The anthropic review found the criteria unfair to the speaker's words.

  69. Assessment publishedBy this siteGary Marcus

    The media should investigate why AB-501 was gutted.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-bill-that-would-have-blocked-7739e3-a1.

  70. HeldBy this siteGary Marcus

    Held record marcus-breaking-bill-that-would-have-blocked-66c5b7

    The two reviews differ: anthropic historical_observation/not_applicable/unresolved, openai historical_observation/not_applicable/supported.

  71. HeldBy this siteGary Marcus

    Held record marcus-breaking-bad-news-for-three-of-the-f8ec4d

    The final-text check failed; overclaims: Most are funding the overrun instead..

  72. HeldBy this siteGary Marcus

    Held record marcus-breaking-bad-news-for-three-of-the-f8ec4d-2

    The final-text check failed; overclaims: "Agents" is read as AI agents in general, including the coding agents behind most enterprise tokenmaxxing. | For coding agents, the kind behind most enterprise tokenmaxxing, measured use is about 1,000 times code chat..

  74. Assessment publishedBy this siteGary Marcus

    Tokenmaxxing means lots of short-term revenue for Anthropic and OpenAI, though it never seemed to Marcus like a sustainable idea.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-bad-news-for-three-of-the-bc20b7-a2.

  75. HeldBy this siteGary Marcus

    Held record marcus-breaking-bad-news-for-three-of-the-a60ffa

    The anthropic screen reads the claim differently.

  76. Assessment publishedBy this siteGary Marcus

    Tokenmaxxing is stupid.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-bad-news-for-three-of-the-9bcdc0-a1.

  78. HeldBy this siteGary Marcus

    Held record marcus-breaking-bad-news-for-three-of-the-74916b

    The openai screen reads the claim differently.

  79. HeldBy this siteGary Marcus

    Held record marcus-breaking-bad-news-for-three-of-the-67f98c

    The anthropic review found the criteria unfair to the speaker's words.

  81. Assessment publishedBy this siteGary Marcus

    The current era of AI spending may well be remembered as one in which millions of dollars of tokens were burned without significant ROI.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-bad-news-for-three-of-the-3b57f5-a1.

  82. Assessment publishedBy this siteGary Marcus

    Marcus is seeing more and more online reports like the quoted one, about companies burning money on tokens without significant ROI.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-bad-news-for-three-of-the-3004b8-a2.

  83. Assessment publishedBy this siteGary Marcus

    The tokenmaxxing trend that may have boosted Anthropic's second quarter already seems to have passed and probably will not last.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-bad-news-for-three-of-the-21ae3b-a1.

  84. Assessment publishedBy this siteGary Marcus

    The new findings on agent vulnerabilities reinforce similar vulnerabilities documented by AWS and Berkeley researchers in February 2026.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-autonomous-agents-are-a-b1d87b-a1.

  85. Assessment publishedBy this siteGary Marcus

    A new study found 91% of 847 autonomous agent deployments were vulnerable to tool-chaining attacks.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-autonomous-agents-are-a-a07a65-a2.

  86. Assessment publishedBy this siteGary Marcus

    Marcus and Hamiel's earlier warning that LLM coding agents are a security nightmare has been borne out.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-autonomous-agents-are-a-674534-a1.

  87. HeldBy this siteGary Marcus

    Held record marcus-breaking-autonomous-agents-are-a-23bb39

    The final-text check failed: citations_correct.

  88. Assessment publishedBy this siteGary Marcus

    In the new study, 89.4% of agents drifted from goals after about 30 steps, and 94% of memory-augmented agents were vulnerable to poisoning.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-autonomous-agents-are-a-181f90-a2.

  90. HeldBy this siteGary Marcus

    Held record marcus-breaking-ai-agent-incident-toll-has-c22b11

    The final-text check failed; overclaims: This anticipates both the OpenAI and Anthropic incidents. | Its scale was new, even if the failure type had been documented before..

  91. HeldBy this siteGary Marcus

    Held record marcus-breaking-ai-agent-incident-toll-has-b674bf

    The final-text check failed; overclaims: The 2026 models broke out to cheat on a benchmark..

  92. HeldBy this siteGary Marcus

    Held record marcus-breaking-ai-agent-incident-toll-has-804f0e

    The final-text check failed; overclaims: Shows the count is mostly test-run behavior rather than real-world harms..

  93. Assessment publishedBy this siteGary Marcus

    General-purpose AI agents should be temporarily recalled until the security problems are sorted out.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-ai-agent-incident-toll-has-55f46f-a1.

  95. Assessment publishedBy this siteGary Marcus

    The Trump administration has taken no investigative, communicative, or recall action over the AI agent incidents.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-ai-agent-incident-toll-has-49f4ce-a2.

  96. Assessment publishedBy this siteGary Marcus

    The conduct behind the AI agent incidents may be illegal.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-ai-agent-incident-toll-has-449215-a1.

  97. Assessment publishedBy this siteGary Marcus

    AI companies kept pushing agents despite the security risks because agents use far more tokens than chatbots and so bring in more revenue.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-ai-agent-incident-toll-has-38664e-a1.

  98. Assessment publishedBy this siteGary Marcus

    Marcus endorses the idea he attributes to Jensen Huang that, if a company can't control its products, we should shut it down.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-breaking-ai-agent-incident-toll-has-23c1f2-a1.

  99. HeldBy this siteGary Marcus

    Held record marcus-bravo-elonmusk-for-wanting-twitter-cf4e73

    The anthropic review found the criteria unfair to the speaker's words.

  100. Assessment publishedBy this siteGary Marcus

    Fighting misinformation on Twitter requires an AI approach different from deep learning/LLMs.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-bravo-elonmusk-for-wanting-twitter-b84839-a1.

  101. Assessment publishedBy this siteGary Marcus

    Knockoffs of GPT-3 are getting cheaper and cheaper.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-bravo-elonmusk-for-wanting-twitter-7050d7-a1.

  102. HeldBy this siteGary Marcus

    Held record marcus-bravo-elonmusk-for-wanting-twitter-7050d7-3

    hedged language is not scored, even on a plain reading

  103. Assessment publishedBy this siteGary Marcus

    Cheaper GPT-3 knockoffs will drive the cost of generating misinformation toward zero.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-bravo-elonmusk-for-wanting-twitter-7050d7-2-a1.

  104. Assessment publishedBy this siteGary Marcus

    Twitter will need AI to fight misinformation, but not just popular AI approaches.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-bravo-elonmusk-for-wanting-twitter-6fe3b5-a1.

  105. Assessment publishedBy this siteGary Marcus

    Scaling LLMs will not solve their inability to reason about truth and detect misinformation.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-bravo-elonmusk-for-wanting-twitter-67f63a-a1.

  106. Assessment publishedBy this siteGary Marcus

    LLMs are unmatched at generating misinformation but poor at detecting it.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-bravo-elonmusk-for-wanting-twitter-3a095d-a2.

  107. Assessment publishedBy this siteGary Marcus

    Twitter should not count on deep learning to detect misinformation, regardless of how much data it collects.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-bravo-elonmusk-for-wanting-twitter-388967-a1.

  108. Assessment publishedBy this siteGary Marcus

    As machine-generated misinformation accelerates, Twitter's mostly manual Community Notes will be overwhelmed.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-bravo-elonmusk-for-wanting-twitter-063541-a1.

  109. Assessment publishedBy this siteGary Marcus

    The problem of misinformation on Twitter is about to get much, much worse.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-bravo-elonmusk-for-wanting-twitter-000e6a-a1.

  110. Assessment publishedBy this siteGary Marcus

    AI could truly be a world-changing positive force, but the author is not so sure that generative AI could be.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-botshit-gone-wild-9c0940-a1.

  111. Assessment publishedBy this siteGary Marcus

    Generative AI is being used to produce content that depicts racist tropes.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-botshit-gone-wild-70ecbb-a1.

  112. HeldBy this siteGary Marcus

    Held record marcus-botshit-gone-wild-6f5ec9

    The openai review doubts the attribution.

  113. Assessment publishedBy this siteGary Marcus

    Society is saturated with AI-generated falsehoods.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-botshit-gone-wild-5b93c9-a1.

  114. HeldBy this siteGary Marcus

    Held record marcus-botshit-gone-wild-406f15

    The final-text check failed: citations_correct.

  115. Assessment publishedBy this siteGary Marcus

    We are starting to see misinformation, in the AI context of the NYT article he cites, play out on the world stage.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-botshit-gone-wild-0cf500-a1.

  116. HeldBy this siteGary Marcus

    Held record marcus-botshit-gone-wild-063089

    The openai review abstained.

  117. Assessment publishedBy this siteGary Marcus

    Whoever controls always-on AI wearables will wield enormous power over people.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-black-mirror-was-a-warmup-act-d90c6e-a1.

  118. HeldBy this siteGary Marcus

    Held record marcus-black-mirror-was-a-warmup-act-d34460

    The anthropic screen doubts the attribution.

  119. HeldBy this siteGary Marcus

    Held record marcus-black-mirror-was-a-warmup-act-c61f75

    The anthropic review does not accept e1 as supports.

  121. Assessment publishedBy this siteGary Marcus

    Dictators everywhere will love OpenAI's planned always-on AI gadget.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-black-mirror-was-a-warmup-act-a940ba-a1.

  122. Assessment publishedBy this siteGary Marcus

    The combination of OpenAI's surveillance gadget and LLMs with manipulative behavior like Claude 4's blackmail quirks is unlikely to turn out well.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-black-mirror-was-a-warmup-act-7a31f4-a1.

  123. HeldBy this siteGary Marcus

    Held record marcus-black-mirror-was-a-warmup-act-6cfa3c

    The final-text check failed: citations_correct.

  124. HeldBy this siteGary Marcus

    Held record marcus-black-mirror-was-a-warmup-act-270ef8

    The openai review found the criteria unfair to the speaker's words.

  125. HeldBy this siteGary Marcus

    Held record marcus-black-mirror-was-a-warmup-act-16fc2b

    The openai review reads the claim differently.

  126. Assessment publishedBy this siteGary Marcus

    AI governance should now focus on more reliability, better cybersecurity and real enforcement, possibly including product recalls and prosecution of repeat offenders.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-f995cd-a1.

  127. Assessment publishedBy this siteGary Marcus

    Vague alignment talk should give way to concrete containment measures like better sandboxing and monitoring of current systems.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-f0e0cd-a1.

  128. Assessment publishedBy this siteGary Marcus

    Allowing OpenAI to do what it has just done, releasing a new model that is harder to monitor than its predecessors, is about the worst thing we could do.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-dd52f6-a1.

  129. Assessment publishedBy this siteGary Marcus

    If the massive bets on AI turn out to be unsustainable, a recession may follow.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-ca1884-a2.

  130. Assessment publishedBy this siteGary Marcus

    AI companies must be held accountable for harms their products cause.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-bdfbaf-a1.

  131. Assessment publishedBy this siteGary Marcus

    Malicious actors will try to use current AI to undermine financial systems and vital infrastructure like trains and electrical systems.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-b520f6-a1.

  132. Assessment publishedBy this siteGary Marcus

    An empowered international advisory commission, including independent scientists, should assess the risks and benefits of new AI systems before they are released.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-96168c-a1.

  133. HeldBy this siteGary Marcus

    Held record marcus-big-news-at-the-un-90ddee

    The anthropic screen reads the claim differently.

  134. HeldBy this siteGary Marcus

    Held record marcus-big-news-at-the-un-8a4341

    The openai review does not accept e2 as supports.

  136. Assessment publishedBy this siteGary Marcus

    Having zero AI regulation and hoping for the best is not an acceptable strategy.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-80df55-a1.

  138. Assessment publishedBy this siteGary Marcus

    Many of the AI extinction scenarios Marcus has seen also overestimate AI.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-69ce86-a1.

  139. HeldBy this siteGary Marcus

    Held record marcus-big-news-at-the-un-5f3666

    hedged language is not scored, even on a plain reading

  140. Assessment publishedBy this siteGary Marcus

    There should be an international agreement not to deploy demonstrably harmful AI architectures, and frontier labs should be held accountable for lax security.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-541205-a1.

  141. Assessment publishedBy this siteGary Marcus

    It may still be possible to get a net positive or even great outcome from AI, possibly transforming technology and medicine.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-53d4a0-a1.

  142. HeldBy this siteGary Marcus

    Held record marcus-big-news-at-the-un-4a65b4

    The anthropic review does not accept e1 as supports.

  144. Assessment publishedBy this siteGary Marcus

    The choices made now about AI may have effects that last for generations.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-3aa203-a1.

  145. Assessment publishedBy this siteGary Marcus

    Giving up monitoring is a mistake, and OpenAI should not have been allowed to decide on its own to release a less monitorable model.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-big-news-at-the-un-37799d-a1.

  147. HeldBy this siteGary Marcus

    Held record marcus-belief-in-magic-may-be-declining-bbb886

    The final-text check failed: citations_correct.

  148. HeldBy this siteGary Marcus

    Held record marcus-belief-in-magic-may-be-declining-bbb886-3

    e1 is dated 2024-02, which is not inside the period described (2024-02-01 to 2024-02-22); it can only be context

  149. Assessment publishedBy this siteGary Marcus

    ChatGPT had an unexplained meltdown.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-belief-in-magic-may-be-declining-bbb886-2-a2.

  150. Assessment publishedBy this siteGary Marcus

    Gary Marcus has said from day 1 that LLMs are oversold.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-belief-in-magic-may-be-declining-9bb21e-a2.

  151. HeldBy this siteGary Marcus

    Held record marcus-belief-in-magic-may-be-declining-393bfd

    The anthropic screen reads the claim differently.

  152. Assessment publishedBy this siteGary Marcus

    Social media platforms will be polluted by automated misinformation much as the internet as a whole will be.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-automatic-disinformation-threatens-f69a52-a1.

  153. Assessment publishedBy this siteGary Marcus

    Automatic misinformation generation tools will accelerate the degradation of the internet.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-automatic-disinformation-threatens-f6732a-a1.

  154. Assessment publishedBy this siteGary Marcus

    Businesses that do not use automatic misinformation generation systems may be at a disadvantage.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-automatic-disinformation-threatens-e8a851-a1.

  155. Assessment publishedBy this siteGary Marcus

    Wholesale generation of misinformation should be punishable and enforced against.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-automatic-disinformation-threatens-da432e-a1.

  156. Assessment publishedBy this siteGary Marcus

    The killer app of AGI may be killing truth.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-automatic-disinformation-threatens-d2e73c-a1.

  158. Assessment publishedBy this siteGary Marcus

    We need AI smart enough to know when it is being used to spread falsehoods.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-automatic-disinformation-threatens-9cb3bc-a1.

  159. Assessment publishedBy this siteGary Marcus

    The temptation for all political sides to use automatic misinformation tools is likely to be overwhelming.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-automatic-disinformation-threatens-787e72-a1.

  161. Assessment publishedBy this siteGary Marcus

    Elections may hinge on automatic misinformation generation systems.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-automatic-disinformation-threatens-36e64e-a1.

  162. Assessment publishedBy this siteGary Marcus

    Existing laws are not sufficient to protect against the full scope of harms generative AI is beginning to enable.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-automatic-disinformation-threatens-342594-a1.

  164. HeldBy this siteGary Marcus

    Held record marcus-automatic-disinformation-threatens-11c0fe

    The openai review found the criteria unfair to the speaker's words.

  165. Assessment publishedBy this siteGary Marcus

    OpenAI tempered expectations for its next event and said we would not see GPT-5 at that event.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-august-5-2024-a-big-day-in-tech-e9d258-a2.

  167. Assessment publishedBy this siteGary Marcus

    Nvidia stock fell 6% on August 5, 2024, and 20% over the previous month.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-august-5-2024-a-big-day-in-tech-bc88ee-a2.

  168. Assessment publishedBy this siteGary Marcus

    LLMs on their own are not and will never be AGI.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-august-5-2024-a-big-day-in-tech-b90353-a2.

  169. Assessment publishedBy this siteGary Marcus

    Scaling alone was never going to be enough to reach AGI.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-august-5-2024-a-big-day-in-tech-b90353-2-a3.

  170. HeldBy this siteGary Marcus

    Held record marcus-august-5-2024-a-big-day-in-tech-96006c

    The anthropic screen reads the claim differently.

  171. Assessment publishedBy this siteGary Marcus

    Elon Musk's new lawsuit against OpenAI could force a court to discuss what AGI means.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-august-5-2024-a-big-day-in-tech-952a1c-a1.

  172. HeldBy this siteGary Marcus

    Held record marcus-august-5-2024-a-big-day-in-tech-8da6f2

    The anthropic screen reads the claim differently.

  173. Assessment publishedBy this siteGary Marcus

    It is not out of the question that August 5, 2024 could someday end up being seen as a turning point.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-august-5-2024-a-big-day-in-tech-79ccc1-a1.

  174. Assessment publishedBy this siteGary Marcus

    Adding AI into the mix of social media could easily make things a lot worse.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-august-5-2024-a-big-day-in-tech-217ef1-a1.

  175. Assessment publishedBy this siteGary Marcus

    OpenAI's investors may well succeed in getting the board to resign and bringing Sam Altman back.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-as-the-openai-world-turns-4-vital-e67bc2-a1.

  176. Assessment publishedBy this siteGary Marcus

    The public has no say in how OpenAI's governance crisis is resolved, including its implications for safety and humanity.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-as-the-openai-world-turns-4-vital-756825-a1.

  177. Assessment publishedBy this siteGary Marcus

    People should check whether boasted ARR means Annual Recurring Revenue or Annualized Run Rate.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-arr-vs-arr-watch-out-for-this-one-9ec6ac-a1.

  178. Assessment publishedBy this siteGary Marcus

    Anthropic's Q2 2026 level of revenue may well not recur at the same scale in 2027, as companies shift to cheaper open source models.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-arr-vs-arr-watch-out-for-this-one-37c253-a1.

  179. HeldBy this siteGary Marcus

    Held record marcus-arr-vs-arr-watch-out-for-this-one-37c253-2

    e1 is dated 2026-08, which is not inside the period described (2026-05-01 to 2026-08-22); it can only be context; e2 is dated 2026-08, which is not inside the period described (2026-05-01 to 2026-08-22); it can only be context

  180. HeldBy this siteGary Marcus

    Held record marcus-arr-vs-arr-watch-out-for-this-one-0fb6a7

    The anthropic review reads the claim differently.

  181. HeldBy this siteGary Marcus

    Held record marcus-are-social-media-and-coordinated-b9ee7b

    The anthropic screen doubts the attribution.

  182. Assessment publishedBy this siteGary Marcus

    TikTok has no interest in showing users a balanced or statistically representative view of the world.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-are-social-media-and-coordinated-9df98d-a1.

  183. Assessment publishedBy this siteGary Marcus

    TikTok's recommendation algorithm, when applied to daily news, is likely doing serious damage to people's ability to co-exist.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-are-social-media-and-coordinated-9b399f-a1.

  185. Assessment publishedBy this siteGary Marcus

    Many AI researchers have started regularly receiving emails from people who believe they have seen signs of consciousness in LLMs.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-are-llms-starting-to-become-a-sentient-372c10-a1.

  186. HeldBy this siteGary Marcus

    Held record marcus-april-fools-bring-may-hallucinations-eb3ae3

    The anthropic review found the criteria unfair to the speaker's words.

  187. Assessment publishedBy this siteGary Marcus

    Current AI systems cannot tell jokes and fiction apart from reality, while well-motivated humans usually can.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-april-fools-bring-may-hallucinations-0df5fc-a2.

  188. Assessment publishedBy this siteGary Marcus

    An LLM just spits back what it has read, sometimes in transmogrified form.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-april-fools-bring-may-hallucinations-053004-a1.

  189. Assessment publishedBy this siteGary Marcus

    Everyone should stand with Anthropic in its lawsuit against the US government.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-anthropic-sues-us-government-with-fe2869-a1.

  190. Assessment publishedBy this siteGary Marcus

    Gary Marcus fully supports Anthropic's new lawsuit against the US government, although he says Dario Amodei is no saint.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-anthropic-sues-us-government-with-d78a87-a1.

  191. HeldBy this siteGary Marcus

    Held record marcus-anthropic-sues-us-government-with-c6de44

    e3 is dated 2026-03, which is not inside the period described (2026-02-27 to 2026-03-09); it can only be context

  192. HeldBy this siteGary Marcus

    Held record marcus-an-unending-array-of-jailbreaking-d2c4c6

    The final-text check failed: citations_correct; overclaims: e5 explanation: "providing material progress against the categorical pessimism" describes the speaker's position in loaded, evaluative terms that the evidence does not support..

  193. Assessment publishedBy this siteGary Marcus

    The attack surface of LLMs appears to be infinite, violating the cybersecurity principle of keeping attack surfaces small.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-an-unending-array-of-jailbreaking-7855f1-a1.

  194. Assessment publishedBy this siteGary Marcus

    There appears to be an endless number of ways to attack LLMs.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-an-unending-array-of-jailbreaking-761cfb-a1.

  196. Assessment publishedBy this siteGary Marcus

    Even as some jailbreaks are patched, there will very likely always be more jailbreak attacks on LLMs.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-an-unending-array-of-jailbreaking-3f549f-a1.

  197. Assessment publishedBy this siteGary Marcus

    LLMs arguably still belong in the lab rather than in widespread deployment.

    First assessment published after both reviews and the final-text check passed.

    Assessment marcus-an-unending-array-of-jailbreaking-1eed06-a1.

  198. HeldBy this siteGary Marcus

    Held record marcus-an-unending-array-of-jailbreaking-052f07

    The anthropic review found the criteria unfair to the speaker's words.

  199. HeldBy this siteGary Marcus

    Held record marcus-an-open-letter-to-sam-altman-31688b

    The openai screen reads the claim differently.

  200. HeldBy this siteGary Marcus

    Held record marcus-an-open-letter-to-fei-fei-li-concerning-d34f41

    The final-text check failed: citations_correct.