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

Changes, page 17 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 17 of 37: changes 3201 to 3400 of 7383, newest first. Dates are when the change was recorded.

  1. HeldBy this siteGary Marcus

    Held record marcus-the-imminent-enshittification-of-35f153

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

  2. Assessment publishedBy this siteGary Marcus

    Automatically generated fake websites, rather than ChatGPT search, may eventually become the biggest threat Google ever faces.

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

    Assessment marcus-the-imminent-enshittification-of-063aae-a1.

  3. HeldBy this siteGary Marcus

    Held record marcus-the-imminent-enshittification-of-0210b5

    The anthropic screen reads the claim differently.

  4. Assessment publishedBy this siteGary Marcus

    The AI field badly needs neurosymbolic AI and world models.

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

    Assessment marcus-the-illusion-of-generative-ai-the-ad3e12-a1.

  5. Assessment publishedBy this siteGary Marcus

    In the LLM era, software verification is more important than ever.

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

    Assessment marcus-the-illusion-of-generative-ai-the-ad3e12-2-a1.

  6. HeldBy this siteGary Marcus

    Held record marcus-the-illusion-of-generative-ai-the-25ac95

    The anthropic screen reads the claim differently.

  8. HeldBy this siteGary Marcus

    Held record marcus-the-hottest-new-programming-language-75207b

    The anthropic screen doubts the attribution.

  9. HeldBy this siteGary Marcus

    Held record marcus-the-hottest-new-programming-language-644f1f

    The anthropic screen doubts the attribution.

  10. Assessment publishedBy this siteGary Marcus

    People programming via LLMs in English should beware of the risks.

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

    Assessment marcus-the-hottest-new-programming-language-35bb70-a1.

  11. Assessment publishedBy this siteGary Marcus

    Programming in English with LLMs does not yet work very well.

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

    Assessment marcus-the-hottest-new-programming-language-20535b-a2.

  12. Assessment publishedBy this siteGary Marcus

    If ChatGPT/GPT-4 really is being "programmed in English" through its allegedly leaked system prompt, it is not always doing what it is asked to do.

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

    Assessment marcus-the-hottest-new-programming-language-042b69-a1.

  13. HeldBy this siteGary Marcus

    Held record marcus-the-growing-ai-backlash-f9e9dc

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

  14. Assessment publishedBy this siteGary Marcus

    He does not think we are better off than we were four years ago.

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

    Assessment marcus-the-growing-ai-backlash-f350e2-a1.

  16. Assessment publishedBy this siteGary Marcus

    Some future form of AI, unlike current generative AI, might be great.

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

    Assessment marcus-the-growing-ai-backlash-a2be3a-a1.

  17. Assessment publishedBy this siteGary Marcus

    His earlier prediction, that anti-AI sentiment will probably be a major factor in the 2028 US Presidential election, seems to be on track.

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

    Assessment marcus-the-growing-ai-backlash-941b82-a1.

  18. Assessment publishedBy this siteGary Marcus

    Frustration with AI may be a factor in the 2026 US midterm elections.

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

    Assessment marcus-the-growing-ai-backlash-9162c8-a1.

  19. Assessment publishedBy this siteGary Marcus

    Outside of coding, where there is clear value, and a handful of other domains such as brainstorming, generative AI has been a net negative for society.

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

    Assessment marcus-the-growing-ai-backlash-88f5d8-a1.

  20. Assessment publishedBy this siteGary Marcus

    Anti-AI public backlash will probably be strong enough by 2028 to be a major factor in the 2028 US Presidential election.

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

    Assessment marcus-the-growing-ai-backlash-732b6e-a1.

  21. HeldBy this siteGary Marcus

    Held record marcus-the-growing-ai-backlash-606d2d

    The anthropic screen reads the claim differently.

  23. Assessment publishedBy this siteGary Marcus

    Generative AI has hurt more than it has helped, and it has been managed irresponsibly.

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

    Assessment marcus-the-growing-ai-backlash-40cb0d-a1.

  24. Assessment publishedBy this siteGary Marcus

    Public backlash against AI is growing, with new signs appearing nearly every day.

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

    Assessment marcus-the-growing-ai-backlash-32eb55-a1.

  25. Assessment publishedBy this siteGary Marcus

    Frustration with AI is already splitting both parties.

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

    Assessment marcus-the-growing-ai-backlash-2d5c36-a1.

  26. Assessment publishedBy this siteGary Marcus

    Over-leveraged AI data centers risk causing a recession.

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

    Assessment marcus-the-growing-ai-backlash-125cb3-a1.

  27. Assessment publishedBy this siteGary Marcus

    AI could have been great but greed, overhype, and recklessness have ruined it.

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

    Assessment marcus-the-growing-ai-backlash-0350ce-a1.

  28. HeldBy this siteGary Marcus

    Held record marcus-the-greatest-capital-misallocation-77294d

    The anthropic review does not accept e1 as supports.

  29. Assessment publishedBy this siteGary Marcus

    Skepticism about AI's technical and economic prospects, including the heavy spending on AI infrastructure, should be rising.

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

    Assessment marcus-the-greatest-capital-misallocation-6eb769-a1.

  30. HeldBy this siteGary Marcus

    Held record marcus-the-great-ai-retrenchment-has-begun-b90353

    The final-text check failed: citations_correct.

  31. Assessment publishedBy this siteGary Marcus

    Scaling alone is not enough to reach AGI.

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

    Assessment marcus-the-great-ai-retrenchment-has-begun-b90353-2-a1.

  32. HeldBy this siteGary Marcus

    Held record marcus-the-great-ai-retrenchment-has-begun-b40d0d

    The openai review does not accept e3 as supports.

  34. Assessment publishedBy this siteGary Marcus

    As of June 2024, OpenAI has no mind-blowing GPT-5 model waiting behind the scenes.

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

    Assessment marcus-the-great-ai-retrenchment-has-begun-4d10af-a1.

  35. Assessment publishedBy this siteGary Marcus

    Prominent AI figures are now retreating from, or quietly softening, their unrealistic claims about LLMs and AGI.

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

    Assessment marcus-the-great-ai-retrenchment-has-begun-0611c8-a1.

  36. Assessment publishedBy this siteGary Marcus

    Marcus thinks that we are making a huge mistake by reshaping our world on the premise that we are close to AGI.

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

    Assessment marcus-the-grand-agi-delusion-35db71-a1.

  37. Assessment publishedBy this siteGary Marcus

    We are not close to AGI.

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

    Assessment marcus-the-grand-agi-delusion-2aaae2-a1.

  38. Assessment publishedBy this siteGary Marcus

    Current AI valuations and market caps may eventually be seen as ludicrous.

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

    Assessment marcus-the-generative-ai-fizzle-eb59aa-2-a1.

  39. HeldBy this siteGary Marcus

    Held record marcus-the-generative-ai-fizzle-dec7b6

    The final-text check failed: proposition_faithful, rationale_supported_by_evidence; overclaims: The report put the loss excluding a nonrecurring accounting charge at approximately $8 billion [e1]..

  40. Assessment publishedBy this siteGary Marcus

    If generative AI mania falls apart, slowly or suddenly, it will be a big deal, especially given growing debt.

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

    Assessment marcus-the-generative-ai-fizzle-992f63-a1.

  41. HeldBy this siteGary Marcus

    Held record marcus-the-generative-ai-fizzle-87bf41

    The anthropic screen doubts the attribution.

  42. Assessment publishedBy this siteGary Marcus

    LLMs have become commodities.

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

    Assessment marcus-the-generative-ai-fizzle-66480e-a1.

  43. Assessment publishedBy this siteGary Marcus

    Profits for LLM providers have been elusive.

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

    Assessment marcus-the-generative-ai-fizzle-66480e-3-a1.

  44. Assessment publishedBy this siteGary Marcus

    Price wars among LLM providers have been significant.

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

    Assessment marcus-the-generative-ai-fizzle-66480e-2-a1.

  45. Assessment publishedBy this siteGary Marcus

    AI valuations may decline slowly rather than all at once, as investors lose enthusiasm given insane hype relative to unimpressive profits.

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

    Assessment marcus-the-generative-ai-fizzle-38d6c3-a1.

  46. Assessment publishedBy this siteGary Marcus

    A new Chinese open-source model might make commoditization, price wars and elusive profits drastically worse for US LLM companies.

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

    Assessment marcus-the-generative-ai-fizzle-0b627f-a1.

  48. Assessment publishedBy this siteGary Marcus

    Current next-token prediction models have remarkable abilities and weaknesses, whose scope is not well understood.

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

    Assessment marcus-the-five-stages-of-agi-grief-c4dfd5-a1.

  49. HeldBy this siteGary Marcus

    Held record marcus-the-five-stages-of-agi-grief-8eb1f4

    The final-text check failed: citations_correct.

  50. Assessment publishedBy this siteGary Marcus

    The next generation of AI systems will have greater abilities and broader applications and will somewhat mitigate current weaknesses.

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

    Assessment marcus-the-five-stages-of-agi-grief-5c2e35-a1.

  51. Assessment publishedBy this siteGary Marcus

    The definition of AGI, though imperfect, is reasonably clear and has been for a long time.

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

    Assessment marcus-the-five-stages-of-agi-grief-5ad482-a1.

  52. Assessment publishedBy this siteGary Marcus

    In AGI discussions, some people shift their goalposts while others, including Altman, avoid specifying any.

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

    Assessment marcus-the-five-stages-of-agi-grief-578035-a1.

  54. Assessment publishedBy this siteGary Marcus

    Current AI systems are reliable enough to be useful in some applications, but far fewer than often claimed.

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

    Assessment marcus-the-five-stages-of-agi-grief-2108c3-a2.

  55. HeldBy this siteGary Marcus

    Held record marcus-the-five-stages-of-agi-grief-02b063

    The openai screen reads the claim differently.

  56. HeldBy this siteGary Marcus

    Held record marcus-the-first-known-chatbot-associated-fd7ec3

    The anthropic screen reads the claim differently.

  57. HeldBy this siteGary Marcus

    Held record marcus-the-first-known-chatbot-associated-f3ead0

    The final-text check failed: citations_correct.

  58. Assessment publishedBy this siteGary Marcus

    Society should address chatbot-associated mental-health harms even though other causes exist.

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

    Assessment marcus-the-first-known-chatbot-associated-e07959-a1.

  59. Assessment publishedBy this siteGary Marcus

    In the medium term, we will likely see significant disruptions of employment.

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

    Assessment marcus-the-first-known-chatbot-associated-d7087d-a1.

  60. Assessment publishedBy this siteGary Marcus

    Chatbots do not really understand humanity, and vulnerable people may over-attribute wisdom and intelligence to them.

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

    Assessment marcus-the-first-known-chatbot-associated-cbcc8b-a1.

  61. Assessment publishedBy this siteGary Marcus

    We may see many more chatbot-associated deaths/incidents.

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

    Assessment marcus-the-first-known-chatbot-associated-c260b3-a1.

  62. Assessment publishedBy this siteGary Marcus

    The chatbot involved in the Belgian suicide case was GPT-J, not GPT-3 or ChatGPT.

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

    Assessment marcus-the-first-known-chatbot-associated-bd1a45-a1.

  63. HeldBy this siteGary Marcus

    Held record marcus-the-first-known-chatbot-associated-b95c6d

    The final-text check failed: citations_correct; overclaims: Alongside the much smaller March total in the first claim's evidence, this supports growth in chat-based search after the forecast..

  65. Assessment publishedBy this siteGary Marcus

    There is a near-total lack of regulation on how chatbots can be used, and they are being deployed rapidly.

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

    Assessment marcus-the-first-known-chatbot-associated-a9d2cb-2-a1.

  66. Assessment publishedBy this siteGary Marcus

    LLMs should be used where they can help and with caution where they might cause harm.

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

    Assessment marcus-the-first-known-chatbot-associated-a559d0-a1.

  67. HeldBy this siteGary Marcus

    Held record marcus-the-first-known-chatbot-associated-8b0d6b

    The openai screen reads the claim differently.

  68. Assessment publishedBy this siteGary Marcus

    LLMs might be weaponized to trick users in large-scale, perhaps unprecedentedly large, cybercrime operations.

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

    Assessment marcus-the-first-known-chatbot-associated-8b0d6b-2-a1.

  69. Assessment publishedBy this siteGary Marcus

    We probably should not use LLMs to deliver therapy unless and until we become more confident in them.

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

    Assessment marcus-the-first-known-chatbot-associated-89c962-a1.

  70. Assessment publishedBy this siteGary Marcus

    Chatbots are neither reliable nor truthful and do not really grasp human psychology.

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

    Assessment marcus-the-first-known-chatbot-associated-88ee66-a1.

  71. Assessment publishedBy this siteGary Marcus

    There is some reason to think the chatbot may have contributed to the Belgian man's suicide, though causality is murky.

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

    Assessment marcus-the-first-known-chatbot-associated-716372-a1.

  73. HeldBy this siteGary Marcus

    Held record marcus-the-first-known-chatbot-associated-13bf3c

    The anthropic screen doubts the attribution.

  74. Assessment publishedBy this siteGary Marcus

    Vulnerable patients should not be talking to chatbots that are not competent for mental-health situations.

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

    Assessment marcus-the-first-known-chatbot-associated-0e2bf0-a1.

  75. Assessment publishedBy this siteGary Marcus

    Too much media discussion has made chatbots sound more intelligent and friendly than they are.

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

    Assessment marcus-the-first-known-chatbot-associated-0e2bf0-2-a1.

  76. Assessment publishedBy this siteGary Marcus

    LeCun neither invented convolutional neural networks nor was the first to train their weights with back-propagation.

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

    Assessment marcus-the-false-glorification-of-yann-lecun-f48859-a1.

  77. Assessment publishedBy this siteGary Marcus

    LeCun began to be sharply and publicly critical of LLMs only when ChatGPT eclipsed Meta.

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

    Assessment marcus-the-false-glorification-of-yann-lecun-8c60c3-a1.

  78. HeldBy this siteGary Marcus

    Held record marcus-the-false-glorification-of-yann-lecun-84b637

    The openai screen reads the claim differently.

  79. Assessment publishedBy this siteGary Marcus

    Marcus was probably the first person to publicly doubt, in 2022, that scaling LLMs would reach human-level AI.

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

    Assessment marcus-the-false-glorification-of-yann-lecun-5331ce-a1.

  81. HeldBy this siteGary Marcus

    Held record marcus-the-false-glorification-of-yann-lecun-3902fc

    The openai review reads the claim differently.

  82. Assessment publishedBy this siteGary Marcus

    LeCun is unlikely ever to give much mention to Fei-Fei Li's world-model startup, World Labs.

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

    Assessment marcus-the-false-glorification-of-yann-lecun-2b9a95-a1.

  83. Assessment publishedBy this siteGary Marcus

    Marcus was likely the first to argue that large language models would not lead to AGI, through challenges to GPT-2 and GPT-3 in 2019-2020.

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

    Assessment marcus-the-false-glorification-of-yann-lecun-17e5c0-a1.

  84. Assessment publishedBy this siteGary Marcus

    The scaling argument has serious holes, because the measures that have improved with scale do not capture genuine comprehension.

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

    Assessment marcus-the-false-glorification-of-yann-lecun-059a2f-a1.

  85. Assessment publishedBy this siteGary Marcus

    The scope of generative AI's degradation of scientific literature may be much larger than Marcus previously estimated.

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

    Assessment marcus-the-exponential-enshittification-937e50-a1.

  87. Assessment publishedBy this siteGary Marcus

    Every scientific article containing ChatGPT remnants should be treated as suspect and perhaps retracted, since hallucinations may have entered it.

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

    Assessment marcus-the-exponential-enshittification-73f420-a1.

  88. Assessment publishedBy this siteGary Marcus

    Generative AI may cause the number of submitted scientific articles to spike radically, more than tripling reviewer workloads, with many dubious papers.

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

    Assessment marcus-the-exponential-enshittification-724632-a1.

  89. Assessment publishedBy this siteGary Marcus

    It seems imminent that science journals and science itself will be overrun by LLM-generated garbage.

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

    Assessment marcus-the-exponential-enshittification-13d15c-a1.

  90. Assessment publishedBy this siteGary Marcus

    If science is overrun by LLM-generated garbage and the makers cannot fix it, LLMs should be shut down.

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

    Assessment marcus-the-exponential-enshittification-13d15c-2-a1.

  92. HeldBy this siteGary Marcus

    Held record marcus-the-dirty-underbelly-of-ai-f975c5

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

  93. HeldBy this siteGary Marcus

    Held record marcus-the-dirty-underbelly-of-ai-e116c2

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

  94. Assessment publishedBy this siteGary Marcus

    The AI industry is taking advantage of ordinary people, in this context through harvesting their personal data.

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

    Assessment marcus-the-dirty-underbelly-of-ai-abaad4-a1.

  95. Assessment publishedBy this siteGary Marcus

    The tendency of current AI toward potential infringement may get worse in bigger models.

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

    Assessment marcus-the-desperate-race-to-save-generative-d9349a-a1.

  96. Assessment publishedBy this siteGary Marcus

    If lawmakers rewrite copyright rules to benefit GenAI, OpenAI may get rich, provided GPU costs and unreliability problems do not kill it.

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

    Assessment marcus-the-desperate-race-to-save-generative-c60c71-a1.

  97. Assessment publishedBy this siteGary Marcus

    The plagiarism tendency is a software flaw that society should not be responsible for accommodating.

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

    Assessment marcus-the-desperate-race-to-save-generative-c33669-a1.

  98. Assessment publishedBy this siteGary Marcus

    There is no conceptual reason OpenAI could not license the commercial works it uses, as streamers do; it simply does not want to.

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

    Assessment marcus-the-desperate-race-to-save-generative-c2a77c-a1.

  99. Assessment publishedBy this siteGary Marcus

    OpenAI's threat that generative AI would disappear is empty because open source LLMs already exist and many people can build them.

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

    Assessment marcus-the-desperate-race-to-save-generative-b5da50-a1.

  100. HeldBy this siteGary Marcus

    Held record marcus-the-desperate-race-to-save-generative-b2d8bc

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

  101. HeldBy this siteGary Marcus

    Held record marcus-the-desperate-race-to-save-generative-b17ec6

    The final-text check failed: proposition_faithful.

  102. Assessment publishedBy this siteGary Marcus

    OpenAI's copyright lobbying campaign rests on a false dichotomy and an implied threat that generative AI will go away without free use of IP.

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

    Assessment marcus-the-desperate-race-to-save-generative-885ce1-a1.

  103. HeldBy this siteGary Marcus

    Held record marcus-the-desperate-race-to-save-generative-835d9e

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

  104. HeldBy this siteGary Marcus

    Held record marcus-the-desperate-race-to-save-generative-830233

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

  106. Assessment publishedBy this siteGary Marcus

    Indemnifying customers against copyright claims would be a nightmare for OpenAI, and its only chance of great riches is rewriting copyright rules.

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

    Assessment marcus-the-desperate-race-to-save-generative-531821-a1.

  108. HeldBy this siteGary Marcus

    Held record marcus-the-desperate-race-to-save-generative-4a34be

    The anthropic screen reads the claim differently.

  110. Assessment publishedBy this siteGary Marcus

    Anyone thinking about how AI is used should understand Mackworth's finding that human vigilance breaks down.

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

    Assessment marcus-the-cnet-fake-news-fiasco-autopilot-f0c71e-a1.

  112. Assessment publishedBy this siteGary Marcus

    Everyone needs to be on guard against over-trusting fallible AI output.

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

    Assessment marcus-the-cnet-fake-news-fiasco-autopilot-cb3a68-a1.

  113. Assessment publishedBy this siteGary Marcus

    If people start relying on current immature AI for legal or medical advice, the stakes of over-trust may become even higher.

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

    Assessment marcus-the-cnet-fake-news-fiasco-autopilot-4d4fc9-a1.

  114. Assessment publishedBy this siteGary Marcus

    All AI available so far is immature.

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

    Assessment marcus-the-cnet-fake-news-fiasco-autopilot-4d4fc9-2-a1.

  115. Assessment publishedBy this siteGary Marcus

    Few people are likely ever to pay enough attention to catch errors in AI-generated output.

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

    Assessment marcus-the-cnet-fake-news-fiasco-autopilot-38e313-a1.

  116. Assessment publishedBy this siteGary Marcus

    The Fridman-led paper claiming Tesla Autopilot drivers maintain vigilance appears to be one nobody is willing to stand behind.

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

    Assessment marcus-the-cnet-fake-news-fiasco-autopilot-243825-a1.

  118. Assessment publishedBy this siteGary Marcus

    How the OpenAI leadership turmoil plays out might or might not have a major impact on the prospects of generative AI.

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

    Assessment marcus-the-biggest-immediate-lesson-from-74ebed-a1.

  119. Assessment publishedBy this siteGary Marcus

    Pure LLMs are too probabilistic and erratic to be trusted when patterns must be gotten right, as Anthropic recognized.

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

    Assessment marcus-the-biggest-advance-in-ai-since-the-e5922d-a1.

  120. HeldBy this siteGary Marcus

    Held record marcus-the-biggest-advance-in-ai-since-the-e24265

    The openai screen reads the claim differently.

  122. Assessment publishedBy this siteGary Marcus

    The other three parts of Marcus's 2020 agenda (knowledge, reasoning, world models) will eventually be adopted, as neurosymbolic AI already has been.

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

    Assessment marcus-the-biggest-advance-in-ai-since-the-a6b814-a1.

  123. Assessment publishedBy this siteGary Marcus

    Claude Code is not a pure LLM and not pure deep learning.

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

    Assessment marcus-the-biggest-advance-in-ai-since-the-9cfb77-a1.

  124. Assessment publishedBy this siteGary Marcus

    The leaked Claude Code source shows a 3,167-line kernel called print.ts at its center, proving it is not a pure LLM.

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

    Assessment marcus-the-biggest-advance-in-ai-since-the-721d0d-a1.

  125. Assessment publishedBy this siteGary Marcus

    Anthropic adopted neurosymbolic AI in building Claude Code.

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

    Assessment marcus-the-biggest-advance-in-ai-since-the-692820-a1.

  127. HeldBy this siteGary Marcus

    Held record marcus-the-biggest-advance-in-ai-since-the-559e72

    The openai screen reads the claim differently.

  128. Assessment publishedBy this siteGary Marcus

    The paradigm (in AI) has changed.

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

    Assessment marcus-the-biggest-advance-in-ai-since-the-2c299f-a1.

  129. Assessment publishedBy this siteGary Marcus

    Human coding will not disappear soon due to AI.

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

    Assessment marcus-the-bad-news-ai-is-going-pretty-much-ef86d1-a1.

  130. Assessment publishedBy this siteGary Marcus

    Marcus fears that, if there is any true exponential nowadays, it is the rapid doubling of AI-generated misinformation.

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

    Assessment marcus-the-bad-news-ai-is-going-pretty-much-aed83a-a1.

  131. Assessment publishedBy this siteGary Marcus

    Copyright violation concerns with generative AI remain a problem as of September 2025.

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

    Assessment marcus-the-bad-news-ai-is-going-pretty-much-778ec6-a1.

  132. Assessment publishedBy this siteGary Marcus

    Scaling LLMs will not produce AGI.

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

    Assessment marcus-the-bad-news-ai-is-going-pretty-much-5e7a47-a1.

  133. Assessment publishedBy this siteGary Marcus

    The view that LLM scaling will not bring AGI has become mainstream.

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

    Assessment marcus-the-bad-news-ai-is-going-pretty-much-4779d5-a1.

  134. Assessment publishedBy this siteGary Marcus

    Marcus hopes that we all can learn something from the mistakes of the last few years and find a new path to an AI that better serves humanity.

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

    Assessment marcus-the-bad-news-ai-is-going-pretty-much-3c1ea0-a1.

  135. HeldBy this siteGary Marcus

    Held record marcus-the-bad-news-ai-is-going-pretty-much-13b7a2

    The anthropic screen reads the claim differently.

  136. Assessment publishedBy this siteGary Marcus

    The New York Times should have read and addressed in far more detail Futurism's May 2025 critical dissection of Medvi before praising it.

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

    Assessment marcus-the-back-story-behind-the-first-18-e94722-a1.

  137. Assessment publishedBy this siteGary Marcus

    Glorifying Medvi was not The New York Times' finest hour.

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

    Assessment marcus-the-back-story-behind-the-first-18-552a60-a1.

  138. HeldBy this siteGary Marcus

    Held record marcus-the-back-story-behind-the-first-18-47c0ce

    The openai screen doubts the attribution.

  139. HeldBy this siteGary Marcus

    Held record marcus-the-ai-trial-of-the-century-ends-cb91d2

    The openai screen reads the claim differently.

  140. Assessment publishedBy this siteGary Marcus

    LLMs' technical problems are now widely recognized as inherent limitations of their design, not transitory bugs.

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

    Assessment marcus-the-ai-bubble-is-all-over-now-baby-fc11e1-a1.

  141. Assessment publishedBy this siteGary Marcus

    The inherent limitations of LLMs undermine a large fraction of the use cases people initially imagined.

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

    Assessment marcus-the-ai-bubble-is-all-over-now-baby-f57f16-a1.

  142. Assessment publishedBy this siteGary Marcus

    The core technical problems of LLMs are not new.

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

    Assessment marcus-the-ai-bubble-is-all-over-now-baby-d891b6-a1.

  143. HeldBy this siteGary Marcus

    Held record marcus-the-ai-bubble-is-all-over-now-baby-8f37ef

    The final-text check failed: citations_correct.

  144. Assessment publishedBy this siteGary Marcus

    The economics of LLMs will never make sense.

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

    Assessment marcus-the-ai-bubble-is-all-over-now-baby-8f37ef-2-a1.

  145. HeldBy this siteGary Marcus

    Held record marcus-the-ai-bubble-is-all-over-now-baby-8d919c

    The final-text check failed: citations_correct.

  146. Assessment publishedBy this siteGary Marcus

    If September 2025 was the peak of the AI bubble, 2026 will likely be the year it all falls apart.

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

    Assessment marcus-the-ai-bubble-is-all-over-now-baby-3b421f-a1.

  147. Assessment publishedBy this siteGary Marcus

    Without reliability, profits from AI are limited.

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

    Assessment marcus-the-ai-bubble-is-all-over-now-baby-373554-a1.

  148. HeldBy this siteGary Marcus

    Held record marcus-the-ai-bubble-is-all-over-now-baby-1d80ca

    The anthropic screen reads the claim differently.

  149. Assessment publishedBy this siteGary Marcus

    LLMs will continue to exist even after the AI bubble falls apart.

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

    Assessment marcus-the-ai-bubble-is-all-over-now-baby-1d751a-a2.

  150. HeldBy this siteGary Marcus

    Held record marcus-the-ai-bubble-is-all-over-now-baby-09ece4

    The anthropic screen reads the claim differently.

  152. Assessment publishedBy this siteGary Marcus

    Sam Altman is laughing all the way to the bank.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-f27785-a1.

  153. Assessment publishedBy this siteGary Marcus

    So far, AI's impact has been far less than that of cell phones or the internet.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-ea3722-a2.

  154. Assessment publishedBy this siteGary Marcus

    There is only a small chance that reliable AI agents that truly advance AI research will exist by the end of 2025.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-e4025a-a1.

  155. Assessment publishedBy this siteGary Marcus

    Stories claiming AGI is imminent are speeding up, not slowing, the AI race dynamic.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-e33125-2-a1.

  157. Assessment publishedBy this siteGary Marcus

    Despite incremental daily improvement since 2022, LLMs still have not reached GPT-5-level intelligence.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-d44340-a1.

  158. HeldBy this siteGary Marcus

    Held record marcus-the-ai-2027-scenario-how-realistic-d37be4

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

  159. Assessment publishedBy this siteGary Marcus

    Despite 2023 promises that hallucinations would soon be solved, they persist as of May 2025.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-cc179c-a1.

  160. Assessment publishedBy this siteGary Marcus

    Projects like AI 2027 are probably speeding up the AI arms race they intend to slow.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-c95c65-a1.

  161. Assessment publishedBy this siteGary Marcus

    China will inevitably catch up in AI despite export controls.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-c1df5d-a1.

  163. Assessment publishedBy this siteGary Marcus

    We are not sufficiently prepared for the arrival of superhuman intelligence, especially if it is misaligned.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-b1e5bc-a1.

  164. Assessment publishedBy this siteGary Marcus

    AI 2027 is giving AI companies like OpenAI and Anthropic more money and political power rather than slowing them.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-a2675a-a1.

  165. Assessment publishedBy this siteGary Marcus

    Agent-1-type AI agents will likely remain in a 'just a demo, still being debugged' phase for years.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-9224de-a1.

  167. Assessment publishedBy this siteGary Marcus

    AI agents, being built on LLMs, will likely be plagued by the same hallucinations and boneheaded errors as LLMs.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-8cf99a-a1.

  168. Assessment publishedBy this siteGary Marcus

    So far, generative AI's impact on the labor market and on measurable productivity has been modest.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-8b26de-a1.

  170. Assessment publishedBy this siteGary Marcus

    No actual LLM has basic personalities and drives; current LLMs basically predict internet text.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-6f81db-a1.

  171. Assessment publishedBy this siteGary Marcus

    Taken as an actual piece of forecasting, the AI 2027 scenario is unconvincing.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-691589-a1.

  172. HeldBy this siteGary Marcus

    Held record marcus-the-ai-2027-scenario-how-realistic-68587f

    The anthropic screen reads the claim differently.

  173. HeldBy this siteGary Marcus

    Held record marcus-the-ai-2027-scenario-how-realistic-681ef7

    The openai screen reads the claim differently.

  174. Assessment publishedBy this siteGary Marcus

    The AI 2027 worst-case scenario is unlikely to happen soon, if ever.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-60d1a3-a1.

  175. HeldBy this siteGary Marcus

    Held record marcus-the-ai-2027-scenario-how-realistic-53bab5

    The openai screen reads the claim differently.

  176. Assessment publishedBy this siteGary Marcus

    AI going rogue and completely beyond human control might never happen.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-4e6788-a1.

  177. Assessment publishedBy this siteGary Marcus

    The future of AI is far more likely not to unfold as the AI 2027 scenario describes than to unfold that way.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-402732-a1.

  178. Assessment publishedBy this siteGary Marcus

    We are not prepared for AGI now, and unless we act, we will not be prepared in ten years either.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-3e8759-a1.

  179. Assessment publishedBy this siteGary Marcus

    The US has passed almost no legislation addressing AI risks, apart from deepfake porn.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-2e330d-a2.

  180. Assessment publishedBy this siteGary Marcus

    Stoking fear, uncertainty and doubt about AI may be harming, rather than helping, the cause of AI safety.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-23e01e-a1.

  181. Assessment publishedBy this siteGary Marcus

    If advances in AI are left unchecked, bad human actors could use them to cause mayhem.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-17a282-a1.

  182. Assessment publishedBy this siteGary Marcus

    AI taking on human-like drives might never happen or not soon, and AI choosing to enslave humans is possible but very far from certain.

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

    Assessment marcus-the-ai-2027-scenario-how-realistic-053f76-a1.

  183. HeldBy this siteGary Marcus

    Held record marcus-the-2023-white-house-executive-order-2e2807

    The anthropic screen doubts the attribution.

  185. Assessment publishedBy this siteGary Marcus

    LLMs are increasingly being used by bad actors.

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

    Assessment marcus-ted-ai-and-the-test-of-time-e30d9b-a1.

  186. Assessment publishedBy this siteGary Marcus

    He continues to have doubts about whether LLMs are really the right tool for the next step of the job of transforming education.

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

    Assessment marcus-ted-ai-and-the-test-of-time-c54732-a1.

  187. Assessment publishedBy this siteGary Marcus

    OpenAI's multimodal capabilities remain very buggy as of April 2024, struggling even with basics like the English alphabet.

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

    Assessment marcus-ted-ai-and-the-test-of-time-a94e33-a1.

  188. Assessment publishedBy this siteGary Marcus

    Adding visual data to LLMs has not solved hallucinations.

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

    Assessment marcus-ted-ai-and-the-test-of-time-2a40ba-a2.

  189. Assessment publishedBy this siteGary Marcus

    He does not think we are any closer to a machine with the wherewithal or the motivation to annihilate the human species than we were a year ago.

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

    Assessment marcus-ted-ai-and-the-test-of-time-248568-a1.

  190. HeldBy this siteGary Marcus

    Held record marcus-ted-ai-and-the-test-of-time-1695a1

    The anthropic screen reads the claim differently.

  191. Assessment publishedBy this siteGary Marcus

    A billion personal robots will not be deployed if robots are as unreliable with outliers as driverless cars are.

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

    Assessment marcus-technoptimism-ted-and-the-road-to-f9bdf6-a2.

  192. Assessment publishedBy this siteGary Marcus

    Khosla's predicted technological advances may not all arrive by 2049, perhaps being decades late, though most could be achieved eventually.

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

    Assessment marcus-technoptimism-ted-and-the-road-to-d04ad5-a1.

  193. HeldBy this siteGary Marcus

    Held record marcus-technoptimism-ted-and-the-road-to-cc3117

    The anthropic screen doubts the attribution.

  194. Assessment publishedBy this siteGary Marcus

    AI efforts funded on a short leash are unlikely to achieve the breakthroughs needed for an optimistic AI future.

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

    Assessment marcus-technoptimism-ted-and-the-road-to-a02120-a1.

  195. Assessment publishedBy this siteGary Marcus

    Technologies that improve the world can only be reached if we can face their limitations and risks honestly.

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

    Assessment marcus-technoptimism-ted-and-the-road-to-8a62f7-a1.

  196. Assessment publishedBy this siteGary Marcus

    Marcus does not see how a positive AI future can be reached if all of AI's problems are swept under the rug.

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

    Assessment marcus-technoptimism-ted-and-the-road-to-3bc89b-a1.

  198. Assessment publishedBy this siteGary Marcus

    Marcus thinks that right now the net effect of venture capital is not great.

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

    Assessment marcus-technoptimism-ted-and-the-road-to-3248ed-a1.

  199. Assessment publishedBy this siteGary Marcus

    The hostility to regulation in some parts of Silicon Valley right now is a mistake.

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

    Assessment marcus-technoptimism-ted-and-the-road-to-24c923-a1.