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

Changes, page 11 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 11 of 55: changes 2001 to 2200 of 10984, newest first. Dates are when the change was recorded.

  1. Assessment publishedBy this siteGary Marcus

    Few people want to fund alternative architectures that can actually obey our instructions, when the immediate money lies in LLMs.

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

    Assessment marcus-llms-dishonest-unpredictable-and-d1de0e-a1.

  2. Assessment publishedBy this siteGary Marcus

    Even models like o3 that can read web links may do better, but confabulation remains widespread across LLMs.

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

    Assessment marcus-llms-dishonest-unpredictable-and-d138ff-a1.

  3. Assessment publishedBy this siteGary Marcus

    No current LLM system can be trusted.

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

    Assessment marcus-llms-dishonest-unpredictable-and-ce086f-a1.

  4. HeldBy this siteGary Marcus

    Held record marcus-llms-dishonest-unpredictable-and-9dec2f

    e1 is dated 2025-06, which is not inside the period described (2025-04-01 to 2025-06-22); it can only be context

  5. HeldBy this siteGary Marcus

    Held record marcus-llms-dishonest-unpredictable-and-776d17

    e1 is dated 2025-06, which is not inside the period described (2025-06-01 to 2025-06-22); it can only be context

  6. HeldBy this siteGary Marcus

    Held record marcus-llms-dishonest-unpredictable-and-538842

    e1 is dated 2025-06, which is not inside the period described (2025-06-01 to 2025-06-22); it can only be context

  7. Assessment publishedBy this siteGary Marcus

    Claude, like LLMs from the start, doesn't understand or reliably follow its instructions, and this is not improving.

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

    Assessment marcus-llms-dishonest-unpredictable-and-4b236d-a1.

  8. HeldBy this siteGary Marcus

    Held record marcus-llms-dishonest-unpredictable-and-46b840

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

  9. HeldBy this siteGary Marcus

    Held record marcus-llms-coding-agents-security-nightmare-d37b11

    hedged language is not scored, even on a plain reading

  11. Assessment publishedBy this siteGary Marcus

    High agency combined with easily manipulated LLM software leads to chaos.

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

    Assessment marcus-llms-coding-agents-security-nightmare-64e0b0-a1.

  14. Assessment publishedBy this siteGary Marcus

    Handling negation words like 'no' and 'not' is a major weak spot for current neural networks.

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

    Assessment marcus-learning-language-is-harder-than-50873c-a2.

  15. Assessment publishedBy this siteGary Marcus

    Systems like GPT-3 are excellent at memorization but are stochastic imitations of their input with very little coherence or comprehension.

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

    Assessment marcus-learning-language-is-harder-than-35a722-a1.

  16. HeldBy this siteGary Marcus

    Held record marcus-large-language-models-like-chatgpt-e27e55

    The final-text check failed: proposition_faithful.

  19. Assessment publishedBy this siteGary Marcus

    The scientific community has no access to whatever error data OpenAI collects on ChatGPT.

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

    Assessment marcus-large-language-models-like-chatgpt-3df6dc-a1.

  20. Assessment publishedBy this siteGary Marcus

    ChatGPT could not reliably reason about the physical world, though it could sometimes produce correct outputs.

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

    Assessment marcus-large-language-models-like-chatgpt-24e93f-a1.

  21. HeldBy this siteGary Marcus

    Held record marcus-kara-swisher-sam-altman-and-the-openai-b40825

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

  22. HeldBy this siteGary Marcus

    Held record marcus-kara-swisher-sam-altman-and-the-openai-a12162

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

  23. HeldBy this siteGary Marcus

    Held record marcus-kara-swisher-sam-altman-and-the-openai-96e3a6

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

  24. HeldBy this siteGary Marcus

    Held record marcus-kara-swisher-sam-altman-and-the-openai-0b5e67

    e1 is dated 2023, which is not inside the period described (… to 2023-11-17); it can only be context; e2 is dated 2023, which is not inside the period described (… to 2023-11-17); it can only be context

  25. Assessment publishedBy this siteGary Marcus

    Within a few months before July 2023, the push to regulate AI, possibly internationally, went from barely being discussed to heated.

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

    Assessment marcus-jumpstarting-ai-governance-4928aa-a1.

  27. HeldBy this siteGary Marcus

    Held record marcus-jumpstarting-ai-governance-1e3802

    The anthropic review does not accept e4 as supports.

  28. HeldBy this siteGary Marcus

    Held record marcus-jumpstarting-ai-governance-1e3802-2

    a observation claim cannot be assessed on the causal_assessment axis

  29. Assessment publishedBy this siteGary Marcus

    The Pandora's Box of LLMs has already been opened.

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

    Assessment marcus-jane-goodall-ai-skeptic-d9919e-a1.

  30. HeldBy this siteGary Marcus

    Held record marcus-jane-goodall-ai-skeptic-d9919e-2

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

  32. HeldBy this siteGary Marcus

    Held record marcus-is-it-time-to-hit-the-pause-button-de9bc1

    The final-text check failed: citations_correct.

  33. HeldBy this siteGary Marcus

    Held record marcus-is-it-time-to-hit-the-pause-button-c79d72

    The anthropic review does not accept e1 as supports.

  34. HeldBy this siteGary Marcus

    Held record marcus-is-elon-musk-dumb-cbcc46

    The final-text check failed: citations_correct.

  35. Assessment publishedBy this siteGary Marcus

    Many people, perhaps thousands or more, will die needlessly because of military AI use.

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

    Assessment marcus-is-ai-already-killing-people-by-accident-c73dcd-a1.

  36. Assessment publishedBy this siteGary Marcus

    Some military AI mistakes will cost lives, some many lives.

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

    Assessment marcus-is-ai-already-killing-people-by-accident-b70644-a1.

  37. HeldBy this siteGary Marcus

    Held record marcus-is-ai-already-killing-people-by-accident-a34a4c

    The final-text check failed: citations_correct.

  38. Assessment publishedBy this siteGary Marcus

    LLMs have always been prone to hallucinations and errors.

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

    Assessment marcus-is-agi-the-right-goal-for-ai-b23dd5-a1.

  39. Assessment publishedBy this siteGary Marcus

    AI benchmarks can be gamed, which weakens an AGI definition built on them.

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

    Assessment marcus-is-agi-the-right-goal-for-ai-aaad0e-a1.

  40. Assessment publishedBy this siteGary Marcus

    As the paper notes, current techniques capture only a fraction of the many facets of cognition that ultimately need to be captured.

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

    Assessment marcus-is-agi-the-right-goal-for-ai-5bdbd2-a1.

  41. Assessment publishedBy this siteGary Marcus

    OpenAI takes the intellectual property of artists and writers without permission.

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

    Assessment marcus-irony-irony-laced-with-karma-and-a7e259-a1.

  42. HeldBy this siteGary Marcus

    Held record marcus-irony-irony-laced-with-karma-and-4417e1

    The openai review does not accept e1 as supports.

  43. HeldBy this siteGary Marcus

    Held record marcus-irony-irony-laced-with-karma-and-43b949

    The openai review doubts the attribution.

  44. HeldBy this siteGary Marcus

    Held record marcus-inside-the-heart-of-chatgpts-darkness-e6b413

    The final-text check failed: proposition_faithful.

  45. Assessment publishedBy this siteGary Marcus

    ChatGPT or other language models will be used for dangerous purposes including massive-scale misinformation.

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

    Assessment marcus-inside-the-heart-of-chatgpts-darkness-a66621-a1.

  46. Assessment publishedBy this siteGary Marcus

    There is little if any government regulation in place to address ChatGPT's risks.

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

    Assessment marcus-inside-the-heart-of-chatgpts-darkness-94566d-a1.

  47. HeldBy this siteGary Marcus

    Held record marcus-inside-the-heart-of-chatgpts-darkness-7c7df2

    e1 is dated 2023-02, which is not inside the period described (2023-02-01 to 2023-02-11); it can only be context

  48. HeldBy this siteGary Marcus

    Held record marcus-inside-the-heart-of-chatgpts-darkness-52141b

    The final-text check failed: proposition_faithful.

  49. HeldBy this siteGary Marcus

    Held record marcus-inside-the-heart-of-chatgpts-darkness-309cf2

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

  50. Assessment publishedBy this siteGary Marcus

    ChatGPT's apparent control of toxic output is an illusion rather than a real solution.

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

    Assessment marcus-inside-the-heart-of-chatgpts-darkness-27ccf8-a1.

  52. HeldBy this siteGary Marcus

    Held record marcus-inside-the-heart-of-chatgpts-darkness-065227

    e1 is dated 2023-02, which is not inside the period described (2023-02-01 to 2023-02-11); it can only be context; e4 is dated 2023-02, which is not inside the period described (2023-02-01 to 2023-02-11); it can only be context

  53. Assessment publishedBy this siteGary Marcus

    Multiple studies have shown that most companies do not see a significant return on their AI investment.

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

    Assessment marcus-if-enough-other-companies-report-d8e731-a1.

  54. HeldBy this siteGary Marcus

    Held record marcus-if-enough-other-companies-report-d05035

    e1 is dated 2026-05, which is not inside the period described (2026-05-01 to 2026-05-26); it can only be context

  55. Assessment publishedBy this siteGary Marcus

    Three companies that have not shown profitability are expected to IPO soon at a combined value of roughly four trillion dollars.

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

    Assessment marcus-if-enough-other-companies-report-b19cc6-a1.

  56. HeldBy this siteGary Marcus

    Held record marcus-if-enough-other-companies-report-84fb45

    e1 is dated 2026-05, which is not inside the period described (2026-05-01 to 2026-05-26); it can only be context; e2 is dated 2026-05, which is not inside the period described (2026-05-01 to 2026-05-26); it can only be context

  57. HeldBy this siteGary Marcus

    Held record marcus-humans-versus-machines-the-hallucination-d4af4a

    The anthropic review reads the claim differently.

  59. HeldBy this siteGary Marcus

    Held record marcus-humanitys-oh-shit-ai-moment-6844a5

    The final-text check failed: citations_correct.

  60. Assessment publishedBy this siteGary Marcus

    LLMs cannot reliably check themselves for hallucinations or for whether their actions are harmful.

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

    Assessment marcus-humanitys-oh-shit-ai-moment-56953e-a1.

  61. Assessment publishedBy this siteGary Marcus

    LLMs do not calculate the consequences of their actions at all.

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

    Assessment marcus-humanitys-oh-shit-ai-moment-36db46-a1.

  62. Assessment publishedBy this siteGary Marcus

    LLMs can easily say one thing and do another, especially when deployed as agents.

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

    Assessment marcus-humanitys-oh-shit-ai-moment-2bcca0-a2.

  63. Assessment publishedBy this siteGary Marcus

    Adding symbolic tools dramatically improves Grok 4's performance on Humanity's Last Exam.

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

    Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-f7583d-a2.

  64. Assessment publishedBy this siteGary Marcus

    Code agents inevitably fail on more complex tasks requiring reasoning.

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

    Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-e81e95-a1.

  65. Assessment publishedBy this siteGary Marcus

    The cutting edge of AI is finally tilting towards neurosymbolic AI, quietly and without public acknowledgement.

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

    Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-dcbd3b-a1.

  66. Assessment publishedBy this siteGary Marcus

    Pure LLMs cannot reliably apply algorithms, while LLMs enhanced with symbolic processes often perform better.

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

    Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-bddd3d-a1.

  67. Assessment publishedBy this siteGary Marcus

    Machine learning has changed the world and will continue to do so.

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

    Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-b32438-a1.

  68. Assessment publishedBy this siteGary Marcus

    Neurosymbolic AI is on track to be a major part of AI's future.

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

    Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-5f0a5b-a1.

  69. Assessment publishedBy this siteGary Marcus

    Models like o3 are far better than non-code-interpreter systems on some tasks but not all, and for now o3 is worse than o1 on hallucinations.

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

    Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-5c5278-a1.

  70. Assessment publishedBy this siteGary Marcus

    Company secrecy has greatly impoverished public science on neurosymbolic AI.

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

    Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-46586b-a2.

  71. HeldBy this siteGary Marcus

    Held record marcus-how-o3-and-grok-4-accidentally-vindicated-3662d2

    The anthropic review does not accept e1 as supports.

  72. Assessment publishedBy this siteGary Marcus

    Systems using code interpreter often outperform those without symbolic tools on many tasks, confirming the value of neurosymbolic AI.

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

    Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-1f42d4-a2.

  73. HeldBy this siteGary Marcus

    Held record marcus-how-o3-and-grok-4-accidentally-vindicated-14c800

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

  74. Assessment publishedBy this siteGary Marcus

    Neural networks learn well but generalize poorly, while symbolic systems generalize well but learn poorly.

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

    Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-1116e9-a2.

  75. Assessment publishedBy this siteGary Marcus

    In 2025 neurosymbolic AI has emerged as the leading approach in AI.

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

    Assessment marcus-how-o3-and-grok-4-accidentally-vindicated-0bc5d5-a1.

  76. HeldBy this siteGary Marcus

    Held record marcus-how-new-are-yann-lecuns-new-ideas-058eb1

    The openai review does not accept e2 as contradicts.

  77. Assessment publishedBy this siteGary Marcus

    Alexa has not added significant features in years and seems clumsy and limited compared to GPT-3.

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

    Assessment marcus-how-come-smart-assistants-have-virtually-cb0a46-a1.

  78. Assessment publishedBy this siteGary Marcus

    LLMs are unruly, and nobody knows how to make them refrain 100% of the time from insulting users, giving bad advice, or making stuff up.

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

    Assessment marcus-how-come-smart-assistants-have-virtually-bc7efe-a1.

  79. HeldBy this siteGary Marcus

    Held record marcus-how-come-smart-assistants-have-virtually-3d874e

    hedged language is not scored, even on a plain reading

  80. Assessment publishedBy this siteGary Marcus

    Chance plays a part, but it does not properly explain why GPT mixes brilliance and stupidity.

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

    Assessment marcus-how-come-gpt-can-seem-so-brilliant-b0d5e4-a1.

  81. HeldBy this siteGary Marcus

    Held record marcus-how-come-gpt-can-seem-so-brilliant-ace35b

    e1 is dated 2022, which is not inside the period described (2022-01-01 to 2022-12-01); it can only be context

  82. Assessment publishedBy this siteGary Marcus

    Almost all of GPT's output is fluent and at least vaguely plausible, not random or word salad.

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

    Assessment marcus-how-come-gpt-can-seem-so-brilliant-8af3b5-a1.

  83. Assessment publishedBy this siteGary Marcus

    GPT's reliance on embeddings makes it good at substituting related phrases, and the same tendency often causes it to make mistakes.

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

    Assessment marcus-how-come-gpt-can-seem-so-brilliant-57661c-a1.

  84. Assessment publishedBy this siteGary Marcus

    GPT reliably produces large amounts of fluent, often plausible prose without needing humans to filter it.

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

    Assessment marcus-how-come-gpt-can-seem-so-brilliant-42e83c-a2.

  85. HeldBy this siteGary Marcus

    Held record marcus-how-come-gpt-can-seem-so-brilliant-323d71

    The anthropic review does not accept e1 as supports.

  86. HeldBy this siteGary Marcus

    Held record marcus-how-come-gpt-can-seem-so-brilliant-0085dc

    e2 is dated 2022, which is not inside the period described (2022-01-01 to 2022-12-01); it can only be context

  87. HeldBy this siteGary Marcus

    Regulation is not always bad for technology and can spur innovation.

    The final-text check failed: proposition_faithful, citations_correct.

  88. Assessment publishedBy this siteGary Marcus

    Getting GPT-5 first will not directly bring China or the US to AGI first, because GPT-5 will not bring us to AGI.

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

    Assessment marcus-how-china-might-crush-the-west-in-496e11-a1.

  89. Assessment publishedBy this siteGary Marcus

    Truthfulness is a major gap in current AI systems, shown by hallucinations and defamatory outputs.

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

    Assessment marcus-how-china-might-crush-the-west-in-088b8f-a1.

  90. HeldBy this siteGary Marcus

    Held record marcus-how-agi-is-nigh-doomers-own-goaled-ee7964

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

  92. Assessment publishedBy this siteGary Marcus

    Doomer hype led governments to start using today's manifestly unreliable AI in war.

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

    Assessment marcus-how-agi-is-nigh-doomers-own-goaled-892d3a-a1.

  93. Assessment publishedBy this siteGary Marcus

    Doomer hype made governments covet AI so much that they have taken it by force.

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

    Assessment marcus-how-agi-is-nigh-doomers-own-goaled-2d46d3-a1.

  94. HeldBy this siteGary Marcus

    Held record marcus-hot-take-on-googles-gemini-3-eb3e5e

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

  95. HeldBy this siteGary Marcus

    Held record marcus-hot-take-on-googles-gemini-3-e74859

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

  96. HeldBy this siteGary Marcus

    Held record marcus-hot-take-on-googles-gemini-3-e39337

    The anthropic review does not accept e1 as contradicts.

  97. Assessment publishedBy this siteGary Marcus

    Google achieved better results than competitors with Gemini 3 without Nvidia GPUs, using only its own TPUs.

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

    Assessment marcus-hot-take-on-googles-gemini-3-6fa4d1-a1.

  98. HeldBy this siteGary Marcus

    Held record marcus-hot-take-on-an-ai-catfight-9ad6ed

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

  99. Assessment publishedBy this siteGary Marcus

    China is already catching up with the U.S. on LLMs.

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

    Assessment marcus-hot-take-on-an-ai-catfight-2227f1-a1.

  100. HeldBy this siteGary Marcus

    Held record marcus-hot-take-on-an-ai-catfight-14a585

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

  101. Assessment publishedBy this siteGary Marcus

    On a first attempt in February 2024, the image model produced no image of a horse riding an astronaut for the 'horse rides astronaut' style prompt.

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

    Assessment marcus-horse-rides-astronaut-redux-51c14f-a1.

  102. Assessment publishedBy this siteGary Marcus

    DALL-E's failure stems from inability to cope with noncanonical prompts rather than from limits of its drawing ability.

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

    Assessment marcus-horse-rides-astronaut-redux-372480-a1.

  103. Assessment publishedBy this siteGary Marcus

    DALL-E 2 is easily fooled by complex sentences and words requiring deep language understanding.

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

    Assessment marcus-horse-rides-astronaut-redux-24077b-a1.

  105. HeldBy this siteGary Marcus

    Held record marcus-horse-rides-astronaut-dba2b0

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

  106. HeldBy this siteGary Marcus

    Held record marcus-horse-rides-astronaut-af6c21

    e2 is dated 2022-05, which is not inside the period described (… to 2022-05-28); it can only be context

  107. HeldBy this siteGary Marcus

    Held record marcus-horse-rides-astronaut-2b3e87

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

  108. HeldBy this siteEd Zitron

    Held record wyea-the-era-of-the-business-idiot-2e754e

    The final-text check did not complete (unavailable).

  109. HeldBy this siteEd Zitron

    Held record wyea-the-era-of-the-business-idiot-1c39da

    The final-text check did not complete (unavailable).

  110. HeldBy this siteEd Zitron

    Held record wyea-the-era-of-the-business-idiot-139343

    The anthropic review reads the claim differently.

  111. HeldBy this siteEd Zitron

    Held record wyea-the-era-of-the-business-idiot-0e2b5e

    e1 measures something else (metric Data Center segment revenue is not Data Center segment revenue as share of total revenue); it can only be context unless it is derived with stated inputs; e2 measures something else (metric total revenue is not Data Center segment revenue as share of total revenue); it can only be context unless it is derived with stated inputs

  112. Assessment publishedBy this siteEd Zitron

    Data center projects will get delayed and loan payments will eventually start going unpaid.

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

    Assessment wyea-the-enshittifinancial-crisis-93b69d-a1.

  113. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-90c14d

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

  114. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-7eea6c

    The final-text check did not complete (unavailable).

  115. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-798f21

    the result turns on our interpretation rather than the speaker's words, so it cannot be scored without a stated plain reading; e1 measures something else (metric share of total revenue from 10%-plus direct customers is not revenue from customers each below 10% of total revenue (derived)); it can only be context unless it is derived with stated inputs; e2 measures something else (period 2025-01-27..2025-04-27 is not 2025-01-27..2025-10-26); it can only be context unless it is derived with stated inputs; e3 measures something else (period 2025-07-28..2025-10-26 is not 2025-01-27..2025-10-26); it

  116. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-74ebcf

    The openai review reads the claim differently.

  117. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-71862a

    e1 is decisive for a financial claim but states no metric scope; the contract's accounting basis is unknown; a financial claim is scored only on a stated basis

  118. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-62aa64

    The final-text check did not complete (unavailable).

  119. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-5fc43a

    The openai review reads the claim differently.

  120. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-5d861d

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

  121. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-5a9e44

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

  122. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-25061a

    The final-text check did not complete (unavailable).

  123. HeldBy this siteEd Zitron

    Held record wyea-the-enshittifinancial-crisis-14396e

    The openai review reads the claim differently.

  124. HeldBy this siteEd Zitron

    Held record wyea-the-cult-of-microsoft-efb30a

    The final-text check did not complete (unavailable).

  125. HeldBy this siteEd Zitron

    Held record wyea-the-cult-of-microsoft-c816c4

    The final-text check did not complete (unavailable).

  126. HeldBy this siteEd Zitron

    Held record wyea-the-cult-of-microsoft-0c1a30

    The final-text check did not complete (unavailable).

  127. HeldBy this siteEd Zitron

    Held record wyea-the-cult-of-microsoft-03ea90

    The final-text check did not complete (unavailable).

  128. HeldBy this siteEd Zitron

    OpenAI is committed to over a trillion dollars of spending with no clear explanation of how it will pay.

    The final-text check failed: citations_correct.

  129. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-fc90c5

    The final-text check did not complete (unavailable).

  130. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-fba95c

    The final-text check did not complete (unavailable).

  131. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-f09ec7

    The openai review reads the claim differently.

  132. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-e90366

    The final-text check did not complete (unavailable).

  133. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-e778e6

    The final-text check did not complete (unavailable).

  134. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-de9e56

    The final-text check did not complete (unavailable).

  135. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-de9e15

    The final-text check did not complete (unavailable).

  136. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-cf5032

    The final-text check failed: rationale_supported_by_evidence; overclaims: Reliability varies by task, tool, and how much human review is involved..

  137. Assessment publishedBy this siteEd Zitron

    Large language models cannot really replace coders.

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

    Assessment wyea-the-case-against-generative-ai-c5d9b3-a1.

  138. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-b947a6

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

  139. Assessment publishedBy this siteEd Zitron

    LLM hallucinations multiply with task complexity and multi-step tasks, which is fatal to the idea of agents.

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

    Assessment wyea-the-case-against-generative-ai-b42d05-a1.

  140. Assessment publishedBy this siteEd Zitron

    Regardless of added reasoning, LLMs cannot be trusted to do things correctly or consistently every time.

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

    Assessment wyea-the-case-against-generative-ai-a9057b-a1.

  141. Assessment publishedBy this siteEd Zitron

    Large language models cannot be relied on to do what users want.

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

    Assessment wyea-the-case-against-generative-ai-a42fdf-a1.

  142. Assessment publishedBy this siteEd Zitron

    There is still no tangible proof that AI is replacing software engineers because AI can't replace them.

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

    Assessment wyea-the-case-against-generative-ai-9d461c-a1.

  143. HeldBy this siteEd Zitron

    Microsoft cannot get even 3% of its 440 million Microsoft 365 subscribers to pay for Microsoft 365 Copilot.

    The final-text check failed: rationale_supported_by_evidence; overclaims: The underlying internal sales materials were not available..

  144. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-8252a5

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

  145. Assessment publishedBy this siteEd Zitron

    OpenAI's Sora cannot replace actors or animators.

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

    Assessment wyea-the-case-against-generative-ai-816b7b-a1.

  146. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-7d6ff9

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

  147. Assessment publishedBy this siteEd Zitron

    Generative AI cannot do a software engineer's job, and it fails while costing huge amounts.

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

    Assessment wyea-the-case-against-generative-ai-7c2756-a1.

  148. Assessment publishedBy this siteEd Zitron

    Coding LLMs do not and never will replace software engineers because LLMs are inherently unreliable.

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

    Assessment wyea-the-case-against-generative-ai-773a61-a1.

  149. Assessment publishedBy this siteEd Zitron

    He sees no future for the generative AI industry beyond a fiery apocalypse.

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

    Assessment wyea-the-case-against-generative-ai-6e6e9c-a1.

  150. Assessment publishedBy this siteEd Zitron

    Coding LLMs do not speed up engineers and actually make them slower.

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

    Assessment wyea-the-case-against-generative-ai-64900a-a1.

  151. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-561c8b

    e2 is dated 2025-09, which is not inside the period described (2025-09-01 to 2025-09-28); it can only be context

  152. HeldBy this siteEd Zitron

    Coding LLMs can generate code but break down and hallucinate more as tasks get more complex.

    The final-text check failed: rationale_supported_by_evidence; overclaims: The located difficulty benchmarks address task success, a different outcome..

  153. Assessment publishedBy this siteEd Zitron

    Demand for generative AI is not there and will never arrive.

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

    Assessment wyea-the-case-against-generative-ai-40e485-a1.

  155. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-3d067c

    e1 measures something else (metric net loss is not cash burn and profitability (net/operating loss)); it can only be context unless it is derived with stated inputs; e2 measures something else (metric net cash used in operating and investing activities is not cash burn and profitability (net/operating loss)); it can only be context unless it is derived with stated inputs; e3 measures something else (metric operating loss is not cash burn and profitability (net/operating loss)); it can only be context unless it is derived with stated inputs

  156. Assessment publishedBy this siteEd Zitron

    AI businesses have no profit lever because model access costs are high and rising.

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

    Assessment wyea-the-case-against-generative-ai-2a33b7-a1.

  157. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-26cf28

    a observation claim cannot be assessed on the causal_assessment axis

  158. Assessment publishedBy this siteEd Zitron

    Under the current deal NVIDIA will not send OpenAI $100 billion, because 90% depends on building 9-10GW of capacity.

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

    Assessment wyea-the-case-against-generative-ai-2231e3-a1.

  159. Assessment publishedBy this siteEd Zitron

    AI data center GPUs have very few use cases outside AI, none of which drive hyperscale revenue, unlike fiber.

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

    Assessment wyea-the-case-against-generative-ai-196074-a1.

  160. HeldBy this siteEd Zitron

    Held record wyea-the-case-against-generative-ai-1728f6

    hedged language is not scored, even on a plain reading

  161. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-f31632

    The final-text check failed: proposition_faithful.

  162. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-e8fce9

    Both reviews say forecast_outcome/conclusive_event/contradicted; the drafted assessment says forecast_outcome/conclusive_event/mixed.

  163. Assessment publishedBy this siteEd Zitron

    SaaS and the business models of venture capital and private equity are collapsing.

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

    Assessment wyea-the-beginning-of-history-e5a6f3-a1.

  164. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-e570b7

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

  165. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-dbdb1e

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

  166. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-dbdb1e-2

    The openai review reads the claim differently.

  167. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-c90ddd

    hedged language is not scored, even on a plain reading

  168. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-a7f4c9

    the conclusive event is after the evidence cutoff; e1 measures something else (company Meta is not Oracle and Meta (including Meta data-center financing vehicles)); it can only be context unless it is derived with stated inputs; e2 is decisive for a financial claim but states no metric scope; e3 is decisive for a financial claim but states no metric scope

  169. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-91cf03

    hedged language is not scored, even on a plain reading

  170. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-7af044

    The anthropic review does not accept e2 as contradicts.

  172. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-57bb6c

    The final-text check did not complete (unavailable).

  173. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-48a8da

    e3 measures something else (company Meta (El Paso data-center financing vehicle) is not AI-related borrowers (Oracle, Meta data-center financing vehicles)); it can only be context unless it is derived with stated inputs; e4 measures something else (company Large tech AI borrowers (Meta, Google, Amazon, Microsoft) is not AI-related borrowers (Oracle, Meta data-center financing vehicles)); it can only be context unless it is derived with stated inputs; e5 measures something else (company Oracle is not AI-related borrowers (Oracle, Meta data-center financing vehicles)); it can only be context unl

  174. HeldBy this siteEd Zitron

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

    The final-text check failed: rationale_supported_by_evidence; overclaims: Crusoe had promised all eight buildings and 1.2 GW by mid-2026, but Oracle said only about 200MW was operating on April 22, and only 42% of capacity was delivered in June. | About two months before the mid-2026 target, only about 200MW of a planned 1.2 GW was operating, by Oracle's own account. | 'Massively' delayed is a judgment call; I treat 42% delivered at the target date as a large delay..

  175. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-394ff3

    The anthropic review does not accept e1 as contradicts.

  176. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-25250e

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

  178. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-11d8f4

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

  179. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-0d4894

    hedged language is not scored, even on a plain reading

  180. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-0cf31c

    The final-text check failed: rationale_supported_by_evidence, citations_correct; overclaims: Some later reporting traces back to documents obtained by Zitron, although it also reports independent Financial Times verification..

  181. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-028190

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

  182. HeldBy this siteEd Zitron

    Held record wyea-the-beginning-of-history-0003c4

    e1 measures something else (company Meta, Alphabet, Microsoft, Amazon, Oracle is not AI industry (hyperscalers and AI labs)); it can only be context unless it is derived with stated inputs; e2 measures something else (company Meta, Alphabet, Microsoft, Amazon, Oracle is not AI industry (hyperscalers and AI labs)); it can only be context unless it is derived with stated inputs; e3 measures something else (company OpenAI is not AI industry (hyperscalers and AI labs)); it can only be context unless it is derived with stated inputs

  184. HeldBy this siteEd Zitron

    Held record wyea-the-anti-economy-e5fa92

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

  185. Assessment publishedBy this siteEd Zitron

    When the author tried to replicate Microsoft's Super Bowl ad prompt, Copilot produced four images with incomprehensible text, only one of which said "Mike's."

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

    Assessment wyea-the-anti-economy-d78f25-a1.

  186. HeldBy this siteEd Zitron

    Held record wyea-the-anti-economy-bd50eb

    e1 is dated 2024-02, which is not inside the period described (2024-01-01 to 2024-02-27); it can only be context

  188. HeldBy this siteEd Zitron

    Held record wyea-the-ai-industry-is-lying-to-you-efe9cb

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

  189. HeldBy this siteEd Zitron

    Held record wyea-the-ai-industry-is-lying-to-you-ea6b1b

    The final-text check failed: citations_correct.

  190. HeldBy this siteEd Zitron

    NVIDIA is earning hundreds of billions of dollars of revenue on chips that are not being installed.

    The final-text check failed: rationale_supported_by_evidence; overclaims: A documented Blackwell deployment also prevents assuming that sold chips generally remained uninstalled..

  191. HeldBy this siteEd Zitron

    Held record wyea-the-ai-industry-is-lying-to-you-d6ab26

    hedged language is not scored, even on a plain reading

  192. HeldBy this siteEd Zitron

    At the current rate, data centers in the announced multi-billion dollar compute deals will not be built before 2030.

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

  193. HeldBy this siteEd Zitron

    At its current construction rate, Stargate Abilene will be fully built in late 2027.

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

  194. HeldBy this siteEd Zitron

    Held record wyea-the-ai-industry-is-lying-to-you-9973f4

    The openai review doubts the attribution.

  195. HeldBy this siteEd Zitron

    Held record wyea-the-ai-industry-is-lying-to-you-8c1207

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

  197. Assessment publishedBy this siteEd Zitron

    Data center delays come from unrealistic construction timelines, not only power, because otherwise more would be finished and waiting for power.

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

    Assessment wyea-the-ai-industry-is-lying-to-you-730283-a1.

  199. Assessment publishedBy this siteEd Zitron

    Stargate Abilene will not be fully completed by October 2026.

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

    Assessment wyea-the-ai-industry-is-lying-to-you-4d4079-a1.

  200. Assessment publishedBy this siteEd Zitron

    AI is not replacing software engineers; they are being laid off because AI tooling is so expensive.

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

    Assessment wyea-the-ai-industry-is-lying-to-you-4bb012-a1.