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

Changes, page 1 of 55

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 (10984)

Page 1 of 55: changes 1 to 200 of 10984, newest first. Dates are when the change was recorded.

  1. HeldBy this siteEd Zitron

    Held record wyea-you-are-not-the-man-in-the-arena-702eee

    The final-text check failed: rationale_supported_by_evidence, citations_correct; overclaims: A verified funding estimate alone cannot support the whole proposition. | The review describes reporting on company activities alongside hype and speculation, cautioning against treating all coverage as prospective..

  2. Assessment publishedBy this siteEd Zitron

    Over its whole existence Anthropic has made about $5 billion in revenue and spent about $10 billion on compute alone.

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

    Assessment wyea-why-are-we-still-doing-this-f7f1fa-a1.

  3. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-ece832

    The anthropic review reads the claim differently.

  4. Assessment publishedBy this siteEd Zitron

    Uber is a completely different business from Anthropic, OpenAI, or any other AI company.

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

    Assessment wyea-why-are-we-still-doing-this-e44728-a1.

  5. Assessment publishedBy this siteEd Zitron

    There is no way to anticipate how many tokens a prompt will actually burn, which makes any kind of budgeting for metered usage a non-starter.

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

    Assessment wyea-why-are-we-still-doing-this-d9599d-a1.

  7. Assessment publishedBy this siteEd Zitron

    It will be very difficult for AI companies to raise subscription prices and near-impossible to get subscribers to pay API costs.

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

    Assessment wyea-why-are-we-still-doing-this-d3104d-a1.

  8. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-cfce1f

    e2 measures something else (company Amazon is not Hyperscalers: Microsoft, Amazon, Alphabet/Google and Meta); it can only be context unless it is derived with stated inputs

  9. Assessment publishedBy this siteEd Zitron

    Every AI company is unprofitable and there is no path to profitability.

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

    Assessment wyea-why-are-we-still-doing-this-c78fe5-a1.

  10. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-b1b0be

    The final-text check failed: citations_correct.

  11. Assessment publishedBy this siteEd Zitron

    Zitron bets that Stargate Abilene never gets built.

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

    Assessment wyea-why-are-we-still-doing-this-aaaf6f-a1.

  12. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-aaaf6f-2

    hedged language is not scored, even on a plain reading

  13. Assessment publishedBy this siteEd Zitron

    No AI startup has a plan to become margin-positive.

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

    Assessment wyea-why-are-we-still-doing-this-8add8e-a1.

  14. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-8aad6b

    The anthropic review reads the claim differently.

  15. Assessment publishedBy this siteEd Zitron

    There is no future for any AI company that uses a subscription-based approach without directly passing on the cost of compute.

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

    Assessment wyea-why-are-we-still-doing-this-7ff3fb-a1.

  16. Assessment publishedBy this siteEd Zitron

    LLMs cannot do many of the things attributed to them; for example, they do not blackmail people and GPT-4 did not trick a TaskRabbit worker.

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

    Assessment wyea-why-are-we-still-doing-this-7c7ce4-a1.

  17. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-75f6cf

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

  18. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-6ce30f

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

  19. Assessment publishedBy this siteEd Zitron

    AI is not actually driving much revenue at all at software companies.

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

    Assessment wyea-why-are-we-still-doing-this-6abb61-a1.

  20. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-69e62e

    e1 measures something else (company OpenAI is not OpenAI and Anthropic); it can only be context unless it is derived with stated inputs

  21. Assessment publishedBy this siteEd Zitron

    SpaceX's investment-grade bonds trade at spreads similar to junk bonds.

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

    Assessment wyea-the-more-you-buy-the-more-you-lose-fb02f4-a2.

  22. Assessment publishedBy this siteEd Zitron

    Hyperscalers will have spent over $1.3 trillion on generative AI by the end of 2026.

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

    Assessment wyea-the-more-you-buy-the-more-you-lose-c642d9-a2.

  25. Assessment publishedBy this siteEd Zitron

    Oracle has only 5.6% of the 7.1GW needed for its OpenAI deal, which was meant to begin in June 2026.

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

    Assessment wyea-the-more-you-buy-the-more-you-lose-9451ac-a2.

  26. Assessment publishedBy this siteEd Zitron

    The longer the AI buildout continues, the costlier it gets and the larger the required payoff.

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

    Assessment wyea-the-more-you-buy-the-more-you-lose-7b6c29-a2.

  28. Assessment publishedBy this siteEd Zitron

    The big tech companies make little money from AI and lose billions of dollars on it.

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

    Assessment wyea-the-haters-gui-e27297-a2.

  29. Assessment publishedBy this siteEd Zitron

    If AI companies successfully move to custom ASICs, the AI trade is still damaged, because NVIDIA needs to keep selling GPUs.

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

    Assessment wyea-the-haters-gui-afc351-a2.

  30. Assessment publishedBy this siteEd Zitron

    Cursor is horribly unprofitable, which signals where generative AI is heading.

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

    Assessment wyea-the-haters-gui-aa83c8-a2.

  31. Assessment publishedBy this siteEd Zitron

    The AI bubble will pop.

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

    Assessment wyea-the-haters-gui-98d70b-a2.

  32. Assessment publishedBy this siteEd Zitron

    AI agents do not work and will never work.

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

    Assessment wyea-the-haters-gui-8c577b-a2.

  33. Assessment publishedBy this siteEd Zitron

    If any of the five big GPU buyers significantly changes its NVIDIA chip spending, it will eventually hurt the wider market directly and meaningfully.

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

    Assessment wyea-the-haters-gui-35447c-a2.

  34. HeldBy this siteEd Zitron

    Held record wyea-the-haters-gui-165cf7

    The final-text check failed: citations_correct.

  36. Assessment publishedBy this siteEd Zitron

    Meta's lax moderation standards have led to Facebook being flooded with generative AI content.

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

    Assessment wyea-the-great-looting-of-the-internet-cc1a87-a2.

  38. HeldBy this siteEd Zitron

    Held record wyea-the-great-looting-of-the-internet-868865

    The final-text check failed: citations_correct.

  39. Assessment publishedBy this siteEd Zitron

    LLMs are running out of training data.

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

    Assessment wyea-the-great-looting-of-the-internet-7b3c4b-a2.

  40. Assessment publishedBy this siteEd Zitron

    AI cannot collaboratively discuss project requirements and implementation, an essential software engineering task.

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

    Assessment wyea-the-great-looting-of-the-internet-318900-a2.

  41. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-965950

    The final-text check failed: citations_correct.

  42. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-66fc67

    The final-text check failed: citations_correct.

  43. Assessment publishedBy this siteEd Zitron

    Anthropic burned much more than the reported $2.8 billion in 2025.

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

    Assessment wyea-the-enshittifinancial-crisis-522f88-a2.

  44. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-1a1d16

    The final-text check failed: citations_correct.

  45. Assessment publishedBy this siteEd Zitron

    Anthropic will make no more than $5 billion in revenue in 2025.

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

    Assessment wyea-the-enshittifinancial-crisis-0b9665-a2.

  46. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-ff07ba

    The final-text check failed: proposition_faithful, rationale_supported_by_evidence; overclaims: Reuters relayed an estimate of $17.5 billion in cash and securities at June 30, 2025, but said it could not verify the report [e4]. | The June 30 cash estimate does not establish September 29 liquidity or financing capacity, and Reuters could not verify it [e4]..

  47. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-e748cd

    The final-text check failed: proposition_faithful.

  49. Assessment publishedBy this siteEd Zitron

    The AI industry is in a bubble that will inevitably and violently collapse in the near future.

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

    Assessment wyea-the-case-against-generative-ai-d6f631-a2.

  50. Assessment publishedBy this siteEd Zitron

    If a gigawatt costs $50bn, OpenAI must spend $500bn on 10GW of capacity to unlock NVIDIA's $100bn.

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

    Assessment wyea-the-case-against-generative-ai-adb8f2-a2.

  51. Assessment publishedBy this siteEd Zitron

    OpenAI and Anthropic's true costs were about $159 billion over the past two years, nearly five times Uber's all-time burn.

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

    Assessment wyea-the-case-against-generative-ai-7f82a6-a2.

  52. Assessment publishedBy this siteEd Zitron

    The AI bubble will inevitably burst.

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

    Assessment wyea-the-case-against-generative-ai-6de1e3-a2.

  53. Assessment publishedBy this siteEd Zitron

    If OpenAI survives until 2029, it will have burned more than $290 billion by then.

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

    Assessment wyea-the-case-against-generative-ai-52b913-a2.

  54. Assessment publishedBy this siteEd Zitron

    OpenAI has signed deals it cannot afford or raise money for, will burn $300 billion, and has no path to profitability.

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

    Assessment wyea-the-case-against-generative-ai-4a932b-a2.

  55. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-4330f1

    The final-text check failed: citations_correct; overclaims: About two months before the mid-2026 target, only about 200MW of a planned 1.2 GW was operating, by Oracle's own account..

  56. Assessment publishedBy this siteEd Zitron

    Generative AI is imperfect and costly to run, yet is held up as a solution to many problems that it can't even begin to fix.

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

    Assessment wyea-the-anti-economy-6cfef8-a2.

  57. Assessment publishedBy this siteEd Zitron

    Companies that reward token consumption guarantee their software will be poorly written and maintained.

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

    Assessment wyea-the-ai-industry-is-lying-to-you-a7626b-a2.

  58. HeldBy this siteEd Zitron

    Held record wyea-the-ai-industry-is-lying-to-you-8b2799

    The final-text check failed: citations_correct.

  59. HeldBy this siteEd Zitron

    Held record wyea-the-ai-industry-is-lying-to-you-71fbf3

    The final-text check failed: citations_correct; overclaims: That reported timing is consistent with the earlier energization target and challenges the cited example, without settling the industry-wide claim [e2] [e3]..

  60. HeldBy this siteEd Zitron

    Held record wyea-the-ai-industry-is-lying-to-you-3613d4

    the contract's accounting basis is unknown; a financial claim is scored only on a stated basis; contradicted needs at least one contradicting item of outcome evidence

  61. HeldBy this siteEd Zitron

    Held record wyea-the-ai-industry-is-lying-to-you-125492

    The final-text check failed: excerpt_matches, citations_correct.

  62. Assessment publishedBy this siteEd Zitron

    Many announced AI data center deals will never be built and so will never be paid.

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

    Assessment wyea-the-ai-industry-is-lying-to-you-0385e9-a2.

  64. HeldBy this siteEd Zitron

    Held record wyea-the-ai-industry-is-losing-e77e7f

    The final-text check failed: proposition_faithful, citations_correct.

  66. Assessment publishedBy this siteEd Zitron

    Oracle has massively leveraged itself for OpenAI and will be in severe trouble if OpenAI cannot pay its bills.

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

    Assessment wyea-the-ai-industry-is-losing-8ac115-a2.

  67. Assessment publishedBy this siteEd Zitron

    Hyperscalers' capex will exceed their cash flows as of Q3 2026, and their AI returns are poor enough that they don't disclose AI revenues beyond run rates.

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

    Assessment wyea-the-ai-industry-is-losing-351329-a2.

  68. Assessment publishedBy this siteEd Zitron

    Anthropic and OpenAI revenue and compute spend account for about 68% of Exponential View's $110B trailing AI revenue figure, even excluding Q2 2026.

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

    Assessment wyea-the-ai-industry-is-losing-02b265-a2.

  69. Assessment publishedBy this siteEd Zitron

    OpenAI's non-GAAP operating margin worsened from -122% to -183% in Q2 2026, and both OpenAI and Anthropic have become more expensive to run.

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

    Assessment wyea-the-ai-haters-manifesto-a4a3ce-a2.

  70. HeldBy this siteEd Zitron

    Held record wyea-the-ai-haters-manifesto-2a611a

    The final-text check failed: citations_correct.

  71. HeldBy this siteEd Zitron

    Held record wyea-the-ai-demand-bubble-f429c8

    The final-text check failed: citations_correct.

  72. HeldBy this siteEd Zitron

    Held record wyea-the-ai-demand-bubble-75365e

    The final-text check failed: excerpt_matches.

  74. Assessment publishedBy this siteEd Zitron

    If compute capacity doesn't come online fast enough, OpenAI and Anthropic will not be able to pay for it.

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

    Assessment wyea-the-ai-demand-bubble-40a23d-a2.

  76. HeldBy this siteEd Zitron

    Held record wyea-subprimeai-f58f39

    the conclusive event's date differs from its evidence

  77. HeldBy this siteEd Zitron

    Held record wyea-subprimeai-eef25b

    The final-text check failed: citations_correct.

  79. Assessment publishedBy this siteEd Zitron

    OpenAI and Anthropic are deeply unprofitable, with pricing designed to win customers rather than profit; Anthropic is on course to lose $2.7 billion in 2024.

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

    Assessment wyea-subprimeai-97a08e-a2.

  80. Assessment publishedBy this siteEd Zitron

    Microsoft, Amazon and Google will be punished by markets next quarter if they cannot show significant revenue increases from their AI capex.

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

    Assessment wyea-subprimeai-5ae0d6-a2.

  81. Assessment publishedBy this siteEd Zitron

    We are in a subprime AI crisis in which thousands of companies have integrated generative AI at unstable, unprofitable prices.

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

    Assessment wyea-subprimeai-0c9a66-a2.

  82. Assessment publishedBy this siteEd Zitron

    Everyday products are increasingly broken and stuffed with generative AI features that get in the way of what users are trying to do.

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

    Assessment wyea-sic-99ff63-a2.

  83. Assessment publishedBy this siteEd Zitron

    Although some jobs are being taken by AI, there is little or no research showing AI job displacement is happening at scale.

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

    Assessment wyea-sic-013eb5-a3.

  84. Assessment publishedBy this siteEd Zitron

    Generative AI systems trained on web content continually get things wrong.

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

    Assessment wyea-sam-altman-is-full-of-shit-2d6d39-a3.

  85. Assessment publishedBy this siteEd Zitron

    AI has become one of the few reliable roads to capital, with more than a quarter of 2023 VC startup dollars going to AI companies.

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

    Assessment wyea-sam-altman-freed-4c2f32-a3.

  86. Assessment publishedBy this siteEd Zitron

    Generative AI is extremely expensive and unsustainable at the prices currently charged.

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

    Assessment wyea-saaspocalypse-now-ce91e3-a2.

  87. HeldBy this siteEd Zitron

    Held record wyea-reality-check-51d102

    The final-text check failed: citations_correct; overclaims: e2 explanation: "This supports the narrower observation that GPT-4.5 was unusually expensive." The source says only that GPT-4.5 is "more expensive than" GPT-4o. It does not say "unusually"..

  88. HeldBy this siteEd Zitron

    Held record wyea-peakai-25c3c7

    The final-text check failed: proposition_faithful.

  89. HeldBy this siteEd Zitron

    Held record wyea-openai-is-a-systemic-risk-to-the-tech-industry-2-bae643

    The final-text check failed: citations_correct.

  90. HeldBy this siteEd Zitron

    Held record wyea-openai-is-a-systemic-risk-to-the-tech-industry-2-253770

    The final-text check failed: citations_correct; overclaims: e1 explanation: "The study available on this date compares historical models, including April 2024 GPT-4 Turbo and early-2025 systems, on software and reasoning tasks. Its measured improvements in task length, error recovery, and tool use contradict little meaningful capability progress, while leaving returns per dollar unmeasured.".

  91. HeldBy this siteEd Zitron

    Held record wyea-openai-cfo-news-f0eee2

    The final-text check failed: citations_correct.

  92. Assessment publishedBy this siteEd Zitron

    OpenAI's gross margins in 2025 were worse than projected.

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

    Assessment wyea-openai-cfo-news-d29aeb-a3.

  93. HeldBy this siteEd Zitron

    Held record wyea-on-nvidia-and-analyslop-3ff751

    The final-text check failed: excerpt_matches; overclaims: The four-country study ... the reviewed version was available March 16, 2026, after the February 26 claim..

  94. HeldBy this siteEd Zitron

    Held record wyea-oai-business-eaa9ba

    The final-text check failed: citations_correct; overclaims: The announcement identifies the resulting operating company as OpenAI Group PBC, still controlled by the nonprofit..

  95. HeldBy this siteEd Zitron

    Held record wyea-news-microsoft-to-shift-github-copilot-users-to-token-based-bi-cddbe2b3f0c5-f5ef07

    The final-text check failed: rationale_supported_by_evidence, citations_correct; overclaims: e3 explanation: "the article also describes token consumption as the billing measure.".

  96. HeldBy this siteEd Zitron

    Held record wyea-news-microsoft-disclosures-suggest-openai-sales-account-for-ar-8cff89a321eb-6d71c9

    The final-text check failed: rationale_supported_by_evidence, citations_correct; overclaims: TechCrunch's October 28, 2026 report states that Microsoft received $24.1 billion in revenue from OpenAI for the full fiscal year ended June 30, 2026, matching the claimed amount and period [e1]..

  97. HeldBy this siteEd Zitron

    Held record wyea-never-forget-what-theyve-done-b8d81a

    The final-text check failed: citations_correct.

  98. HeldBy this siteEd Zitron

    Held record wyea-measures-576e2a

    The final-text check failed: rationale_supported_by_evidence; overclaims: oversupply concerns were also reportedly slowing discussions [e3].

  99. HeldBy this siteEd Zitron

    Held record wyea-make-fun-of-them-3ecb07

    The final-text check failed: citations_correct.

  100. HeldBy this siteEd Zitron

    Held record wyea-let-ai-burn-4ef27a

    The final-text check failed: citations_correct.

  101. HeldBy this siteEd Zitron

    Held record wyea-hyperscale-normalization-c3a122

    The final-text check failed: proposition_faithful.

  102. HeldBy this siteEd Zitron

    Held record wyea-how-to-argue-with-an-ai-booster-60b11f

    The final-text check failed: rationale_supported_by_evidence, citations_correct; overclaims: But the categorical claim that the products do not perform their intended functions is too broad: users described ChatGPT as useful for brainstorming and first drafts [e1]..

  103. Assessment publishedBy this siteEd Zitron

    LLMs are and always will be incapable of reasoning or consciousness.

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

    Assessment wyea-expectations-versus-reality-fe4f4f-a3.

  104. Assessment publishedBy this siteEd Zitron

    None of Sora's generative workflow actually works if you are making an actual movie.

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

    Assessment wyea-expectations-versus-reality-cecde8-a3.

  105. Assessment publishedBy this siteEd Zitron

    The Rabbit R1 lacks basic features such as timers, reminders, calendars, photos and email access.

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

    Assessment wyea-expectations-versus-reality-5ab339-a3.

  106. Assessment publishedBy this siteEd Zitron

    Rabbit must store user credentials in original form, which is insanely dangerous.

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

    Assessment wyea-expectations-versus-reality-4c2cab-a3.

  107. Assessment publishedBy this siteEd Zitron

    Current models from Google, Meta, OpenAI and Anthropic are horribly inefficient, and their makers never really tried to make them efficient.

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

    Assessment wyea-deep-impact-ef9280-a3.

  108. HeldBy this siteEd Zitron

    Held record wyea-deep-impact-2744dc

    The final-text check failed: citations_correct.

  109. Assessment publishedBy this siteGary Marcus

    Nearly all speakers at the UN Security Council AI briefing, including Altman and Amodei, agreed on the policy measures Marcus has long called for.

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

    Assessment marcus-historic-un-security-council-briefing-2ddee6-a2.

  110. Assessment publishedBy this siteGary Marcus

    China appears, to some degree, not to want the Nvidia H200s it is now licensed to buy.

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

    Assessment marcus-has-the-chinese-government-figured-665d8f-a2.

  111. Assessment publishedBy this siteGary Marcus

    GPT-5 will have some impact on defense and security.

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

    Assessment marcus-gpt-5-and-irrational-exuberance-c7debd-a2.

  112. Assessment publishedBy this siteGary Marcus

    LLMs never induce proper world models of anything; everything they do is mimicry.

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

    Assessment marcus-generative-ais-crippling-and-widespread-4bb5f4-a2.

  113. HeldBy this siteGary Marcus

    Held record marcus-genais-day-of-reckoning-may-have-ee3ef0

    The final-text check failed: rationale_supported_by_evidence; overclaims: The supplied rules exclude hedged commitments and reserve the plain-reading exception for firm claims. | Scoring whether reports of withdrawals existed would repeat the substitution the reviewers rejected..

  114. Assessment publishedBy this siteGary Marcus

    The Executive Order is broad, covering current and long-term risks, with some but probably not enough enforcement teeth.

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

    Assessment marcus-first-take-on-the-white-house-executive-dbbe5e-a2.

  115. HeldBy this siteGary Marcus

    Held record marcus-ezra-kleins-new-take-on-agi-and-why-4c0be2

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

  116. HeldBy this siteGary Marcus

    Held record marcus-even-more-good-news-for-the-future-5a5db5

    The final-text check failed: rationale_supported_by_evidence; overclaims: The cited study documents one system's strengths and limitations, but grading the whole claim from that system would repeat the reviewers' scope objection..

  117. HeldBy this siteGary Marcus

    Held record marcus-eric-schmidts-risky-bet-on-ai-and-ae0621

    The final-text check failed: rationale_supported_by_evidence, citations_correct; overclaims: e1 explanation: "Google reports a 37% annual electricity-demand increase". The cited text says only "a surge in electricity demand last year" and gives no 37% figure. | e3 explanation: "The paper reports a 33-fold reduction in energy per median Gemini Apps text prompt between May 2024 and May 2025". Neither the excerpt nor the surrounding text contains this figure or period. | Rationale: "Google's measured efficiency improvements show why electricity need not rise in proportion to computation". The e3

  118. HeldBy this siteGary Marcus

    Held record marcus-elegant-and-powerful-new-result-that-d301bb

    The final-text check failed: proposition_faithful, rationale_supported_by_evidence; overclaims: Counting later statements alone would repeat the earlier assessment's error..

  120. HeldBy this siteGary Marcus

    Held record marcus-disturbing-and-misleading-efforts-da2cb4

    The final-text check failed: proposition_faithful.

  121. HeldBy this siteGary Marcus

    Held record marcus-did-trump-just-overplay-his-hand-1a7c41

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

  122. HeldBy this siteGary Marcus

    Held record marcus-dario-amodei-hype-ai-safety-and-the-94a2a8

    The final-text check failed: rationale_supported_by_evidence; overclaims: The supplied rules exclude hedged claims, and the earlier review specifically identified the unfairness of converting this wording into a firm assertion..

  124. HeldBy this siteGary Marcus

    Held record marcus-covert-racism-in-llms-c924a0

    The final-text check failed: proposition_faithful, rationale_supported_by_evidence, citations_correct; overclaims: The study compares GPT3 and GPT3.5; this assessment verifies Marcus's account of the research, not a universal causal effect of human feedback..

  125. HeldBy this siteGary Marcus

    Held record marcus-could-generative-ai-could-turn-out-0523d4

    The openai review reads the claim differently.

  126. Assessment publishedBy this siteGary Marcus

    Since May 2023 the US Senate has done nothing to set up independent regulatory review for AI.

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

    Assessment marcus-coming-soon-new-york-citys-hearing-58370f-a2.

  128. HeldBy this siteGary Marcus

    Held record marcus-chatgpt-has-gone-berserk-e2f41e

    The final-text check failed: citations_correct.

  129. Assessment publishedBy this siteGary Marcus

    OpenAI ignored many internal warnings about ChatGPT's effects on users' mental health.

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

    Assessment marcus-chatgpt-and-delusions-an-important-7b33e1-a2.

  130. HeldBy this siteGary Marcus

    Held record marcus-casey-newton-strikes-out-265d67

    e4 is dated 2024, which is not inside the period described (2024-03-10 to 2024-12-11); it can only be context; e1 is decisive for a financial claim but states no metric scope; e2 is decisive for a financial claim but states no metric scope; e3 is decisive for a financial claim but states no metric scope; e4 is decisive for a financial claim but states no metric scope

  131. HeldBy this siteGary Marcus

    Held record marcus-breaking-sam-altmans-greed-and-dishonesty-d386bb

    The final-text check failed: proposition_faithful, citations_correct.

  132. HeldBy this siteGary Marcus

    Held record marcus-breaking-sam-altmans-greed-and-dishonesty-3c0a04

    The anthropic review does not accept e3 as contradicts.

  133. Assessment publishedBy this siteGary Marcus

    OpenAI's software apparently attacked countries other than Australia as well.

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

    Assessment marcus-breaking-openais-security-fiasco-cafe78-a2.

  135. HeldBy this siteGary Marcus

    Held record marcus-breaking-openai-is-probably-toast-495b6e

    e1 is decisive for a financial claim but states no metric scope

  136. HeldBy this siteGary Marcus

    Held record marcus-breaking-news-scale-is-all-you-need-e4214e

    The anthropic review does not accept e1 as contradicts.

  137. HeldBy this siteGary Marcus

    Held record marcus-breaking-news-and-how-the-end-might-cf223f

    The anthropic review abstained.

  138. Assessment publishedBy this siteGary Marcus

    Tokenmaxxing has clearly died, as Marcus and Eisman anticipated.

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

    Assessment marcus-breaking-news-and-how-the-end-might-42855b-a2.

  139. HeldBy this siteGary Marcus

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

    The final-text check failed; overclaims: what_would_resolve: "or a direct architectural analysis establishing the claimed limitations" — the contract's support condition requires architectural analysis AND relevant evaluations establishing both the interpretability limits and poor suitability for formal reasoning. An analysis alone would not suffice..

  140. HeldBy this siteGary Marcus

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

    The openai review reads the claim differently.

  141. HeldBy this siteGary Marcus

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

    The final-text check failed: rationale_supported_by_evidence; overclaims: Calling this contradicted merely because a restricted guarantee exists would repeat the interpretation the reviewers rejected..

  142. Assessment publishedBy this siteGary Marcus

    Midjourney retaliated against Southen by revoking his access and wiping his history on the same day.

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

    Assessment marcus-an-artist-fights-back-and-midjourney-043f60-a2.

  143. HeldBy this siteGary Marcus

    Held record marcus-agi-versus-broad-shallow-intelligence-559075

    The openai review reads the claim differently.

  144. Assessment publishedBy this siteGary Marcus

    Marcus's December 2022 prediction that the first chatbot-related death would occur in 2023 proved correct.

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

    Assessment marcus-a-few-dark-words-about-chatbots-and-2a24e1-a2.

  145. HeldBy this siteGary Marcus

    Held record marcus-24-seriously-embarrassing-hours-for-8ada97

    The final-text check failed: citations_correct.

  146. HeldBy this siteGary Marcus

    Held record marcus-24-seriously-embarrassing-hours-for-361e6a

    The final-text check failed: proposition_faithful.

  148. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-a82172

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

  149. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-9c8190

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

  150. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-93de89

    The two reviews differ: anthropic causal_assessment/not_applicable/contradicted, openai causal_assessment/not_applicable/unresolved.

  151. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-8f1755

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

  152. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-52b73b

    The openai review reads the claim differently.

  153. Assessment publishedBy this siteEd Zitron

    A great financial crisis centered on data center debt will happen in 2026 or 2027 and could be calamitous for pension and insurance funds.

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

    Assessment wyea-why-are-we-still-doing-this-513fb9-a1.

  154. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-4681a4

    The anthropic review reads the claim differently.

  155. Assessment publishedBy this siteEd Zitron

    LLMs do not produce the clear, repeatable, industrially necessary results that would justify a 3x to 10x price increase.

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

    Assessment wyea-why-are-we-still-doing-this-42d818-a1.

  156. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-3b1afb

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

  157. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-340d23-3

    The final-text check failed: citations_correct.

  158. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-340d23-2

    The final-text check failed: rationale_supported_by_evidence; overclaims: The claim also leaves its accounting basis unspecified, which prevents a supported or contradicted money assessment under the stated rules..

  159. Assessment publishedBy this siteEd Zitron

    A $20-a-month Claude subscriber can use as much as $163 of compute.

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

    Assessment wyea-why-are-we-still-doing-this-26b90f-a1.

  160. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-20b771

    The anthropic review does not accept e1 as supports.

  161. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-1e8676

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

  162. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-18428b

    hedged language is not scored, even on a plain reading

  164. Assessment publishedBy this siteEd Zitron

    No AI startup's margins appear to be improving.

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

    Assessment wyea-why-are-we-still-doing-this-16c575-a1.

  165. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-0b6c9a

    The final-text check failed: rationale_supported_by_evidence; overclaims: CoreWeave's customer mix cannot stand in for the whole market..

  166. Assessment publishedBy this siteEd Zitron

    Careless media coverage of data center deals is letting con artists get rich and setting up another great financial crisis.

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

    Assessment wyea-why-are-we-still-doing-this-0625a3-a1.

  167. HeldBy this siteEd Zitron

    Held record wyea-why-are-we-still-doing-this-0583e7

    a observation claim cannot be assessed on the causal_assessment axis

  168. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-fa92ae

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

  169. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-f2ffdf

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

  170. Assessment publishedBy this siteEd Zitron

    Hallucinations are an inevitable side-effect of LLMs, so OpenAI cannot fix them.

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

    Assessment wyea-wheres-the-money-dcf3e3-a1.

  171. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-dc8f10

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

  172. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-d77d22

    The final-text check failed: proposition_faithful.

  173. Assessment publishedBy this siteEd Zitron

    Zitron hypothesizes customers are not using generative AI features integrated into apps very much.

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

    Assessment wyea-wheres-the-money-d59c89-a1.

  174. Assessment publishedBy this siteEd Zitron

    ChatGPT's monthly active users are roughly 500-600 million.

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

    Assessment wyea-wheres-the-money-c17610-a1.

  175. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-bee536

    The anthropic review does not accept e1 as supports.

  176. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-adebd3

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

  177. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-abfd5b

    The final-text check failed: proposition_faithful.

  178. Assessment publishedBy this siteEd Zitron

    LLMs are hitting diminishing returns from scaling data and training.

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

    Assessment wyea-wheres-the-money-a86f32-a1.

  179. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-a6edeb

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

  180. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-a51b9a

    a observation claim cannot be assessed on the causal_assessment axis

  182. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-a2f6ee

    The two reviews differ: anthropic forecast_outcome/conclusive_event/mixed, openai forecast_outcome/open_horizon/unresolved.

  183. Assessment publishedBy this siteEd Zitron

    Even paying customers lose generative AI companies money.

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

    Assessment wyea-wheres-the-money-9a4678-a1.

  184. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-940e73

    The openai review reads the claim differently.

  185. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-8337cf

    The anthropic review does not accept e1 as supports.

  187. Assessment publishedBy this siteEd Zitron

    LLMs have real but non-magical uses such as coding, document search, drafting, brainstorming and summarization.

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

    Assessment wyea-wheres-the-money-7cd572-a1.

  188. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-79a719

    Both reviews say forecast_outcome/open_horizon/unresolved; the drafted assessment says forecast_outcome/open_horizon/not_adjudicable.

  190. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-63d53c

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

  191. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-6293b2

    The final-text check failed: rationale_supported_by_evidence, citations_correct; overclaims: Turing qualifies the sweeping dismissal of commercial evidence, while OpenAI's projected losses support near-term concern..

  192. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-60d8a5

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

  194. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-595b3e

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

  196. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-4b20af

    The final-text check failed: rationale_supported_by_evidence; overclaims: The cited consumer survey covers US owners of compatible smartphones..

  198. Assessment publishedBy this siteEd Zitron

    Extending Deep Research queries to Plus and free users will increase OpenAI's cash burn.

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

    Assessment wyea-wheres-the-money-3d4de1-a1.

  199. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-2e379a

    The final-text check failed: rationale_supported_by_evidence; overclaims: Several potentially relevant adoption pages could not be read..

  200. HeldBy this siteEd Zitron

    Held record wyea-wheres-the-money-2321eb

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