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

Changes, page 13 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 13 of 37: changes 2401 to 2600 of 7383, newest first. Dates are when the change was recorded.

  3. Assessment publishedBy this siteGary Marcus

    Newer LLMs still lack genuine comprehension and are therefore still very easy for anyone to break.

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

    Assessment marcus-what-does-it-mean-when-an-ai-fails-d5182b-a2.

  7. Assessment publishedBy this siteGary Marcus

    Current LLMs have no real physical, social, or psychological understanding; they are richer parroting of larger corpora.

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

    Assessment marcus-what-does-it-mean-when-an-ai-fails-b49d13-a1.

  8. Assessment publishedBy this siteGary Marcus

    Marcus sees no reason whatsoever to think that the underlying problem of these models, a lack of cognitive models of the world, has been remedied.

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

    Assessment marcus-what-does-it-mean-when-an-ai-fails-ae0bd9-a1.

  9. Assessment publishedBy this siteGary Marcus

    LLMs on their own, without symbolic hybridization, are inherently weak at understanding the physical and psychological world.

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

    Assessment marcus-what-does-it-mean-when-an-ai-fails-a78bb4-a1.

  11. Assessment publishedBy this siteGary Marcus

    GPT-4 will very likely still have a fundamental lack of reliability, though it may take longer prompts to break.

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

    Assessment marcus-what-does-it-mean-when-an-ai-fails-7fce77-a1.

  12. Assessment publishedBy this siteGary Marcus

    Because LLMs lack causal world models, they are not stable or trustworthy.

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

    Assessment marcus-what-does-it-mean-when-an-ai-fails-739569-a1.

  13. HeldBy this siteGary Marcus

    Held record marcus-what-does-it-mean-when-an-ai-fails-6f4194

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

  14. HeldBy this siteGary Marcus

    Held record marcus-what-does-it-mean-when-an-ai-fails-204104

    The final-text check failed: citations_correct.

  15. Assessment publishedBy this siteGary Marcus

    Scott Alexander's finding that InstructGPT got four of six previously failed prompts right may have been a fluke.

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

    Assessment marcus-what-does-it-mean-when-an-ai-fails-112099-a1.

  16. HeldBy this siteGary Marcus

    Held record marcus-what-did-they-know-and-when-did-they-f05abb

    e3 is dated 2023-02, which is not inside the period described (2022-12-05 to 2023-02-21); it can only be context

  17. HeldBy this siteGary Marcus

    Held record marcus-what-did-they-know-and-when-did-they-d0b3c5

    The anthropic screen reads the claim differently.

  19. HeldBy this siteGary Marcus

    Held record marcus-what-did-they-know-and-when-did-they-bf906c

    The anthropic screen reads the claim differently.

  20. HeldBy this siteGary Marcus

    Held record marcus-what-did-they-know-and-when-did-they-09a79b

    e1 is dated 2023-02, which is not inside the period described (2023-02-01 to 2023-02-21); it can only be context

  21. Assessment publishedBy this siteGary Marcus

    If OpenAI releases LLM-based agents carelessly, everyone will be in grave danger.

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

    Assessment marcus-what-could-possibly-go-wrong-with-d1e83a-a1.

  22. Assessment publishedBy this siteGary Marcus

    At current technological readiness, giving LLMs full control of users' systems is reckless and should not be done.

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

    Assessment marcus-what-could-possibly-go-wrong-with-c2b646-a1.

  23. Assessment publishedBy this siteGary Marcus

    Even if the $7 trillion bet succeeded, it might tear apart society through unemployment, inequality or upheaval in geopolitics.

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

    Assessment marcus-what-could-possibly-go-wrong-with-b65fa9-a1.

  24. Assessment publishedBy this siteGary Marcus

    Marcus does not see LLM-based agents changing the world for the better, which he calls the very best case, and certainly not this decade.

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

    Assessment marcus-what-could-possibly-go-wrong-with-b63ff8-a1.

  25. Assessment publishedBy this siteGary Marcus

    AI agent technology is unlikely to reach the needed maturity for reliable use in the next several years.

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

    Assessment marcus-what-could-possibly-go-wrong-with-b0d7bf-a1.

  26. Assessment publishedBy this siteGary Marcus

    There are signs OpenAI may not be doing very well economically.

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

    Assessment marcus-what-could-possibly-go-wrong-with-aec789-a1.

  27. Assessment publishedBy this siteGary Marcus

    The best realistic outcome is that LLM-based agents never see widespread release, dying because of their likely reliability problems.

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

    Assessment marcus-what-could-possibly-go-wrong-with-93f299-a1.

  28. Assessment publishedBy this siteGary Marcus

    Marcus puts LLM-based agents distributed to millions of customers at the top of his own list of the most dangerous ideas of 2024.

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

    Assessment marcus-what-could-possibly-go-wrong-with-92cd48-a1.

  29. Assessment publishedBy this siteGary Marcus

    Marcus is not convinced that, if Altman's $7 trillion project succeeded, its proceeds would be redistributed remotely equitably.

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

    Assessment marcus-what-could-possibly-go-wrong-with-85735a-a1.

  30. Assessment publishedBy this siteGary Marcus

    If OpenAI's software can take over user devices, rogue employees could take over devices and attackers will have strong incentive to hack OpenAI.

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

    Assessment marcus-what-could-possibly-go-wrong-with-83504e-a1.

  31. HeldBy this siteGary Marcus

    Held record marcus-what-could-possibly-go-wrong-with-796e20

    The openai review reads the claim differently.

  32. Assessment publishedBy this siteGary Marcus

    If LLM agents ran on Defense Department computers, a single LLM error could kill many people.

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

    Assessment marcus-what-could-possibly-go-wrong-with-7817ff-a1.

  34. HeldBy this siteGary Marcus

    Held record marcus-what-could-possibly-go-wrong-with-6b4c1c

    The openai review reads the claim differently.

  35. Assessment publishedBy this siteGary Marcus

    Marcus is not confident that Altman or anyone else has any idea, in the near term, of how to make AGI safe.

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

    Assessment marcus-what-could-possibly-go-wrong-with-668e3f-a1.

  36. HeldBy this siteGary Marcus

    Held record marcus-what-could-possibly-go-wrong-with-637899

    The final-text check failed: citations_correct.

  37. Assessment publishedBy this siteGary Marcus

    The list of OpenAI/LLM security issues will continue to grow.

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

    Assessment marcus-what-could-possibly-go-wrong-with-619211-a1.

  38. HeldBy this siteGary Marcus

    Held record marcus-what-could-possibly-go-wrong-with-5f85bb

    The openai screen doubts the attribution.

  39. Assessment publishedBy this siteGary Marcus

    When LLMs can take actions like writing emails or signing checks, the consequences of their unreliability and hallucinations escalate.

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

    Assessment marcus-what-could-possibly-go-wrong-with-5e2c56-a1.

  40. Assessment publishedBy this siteGary Marcus

    Marcus calls device-controlling LLM agents, given their privacy problems combined with worries about security, 'a colossal accident waiting to happen'.

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

    Assessment marcus-what-could-possibly-go-wrong-with-3ba60a-a1.

  41. Assessment publishedBy this siteGary Marcus

    Whatever numbers investors might run, Marcus is not remotely convinced that we are particularly close to AGI.

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

    Assessment marcus-what-could-possibly-go-wrong-with-2f40c0-a1.

  43. HeldBy this siteGary Marcus

    Held record marcus-what-could-possibly-go-wrong-with-204c35

    The openai review reads the claim differently.

  44. Assessment publishedBy this siteGary Marcus

    AI agents need to be built on a stable and secure foundation.

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

    Assessment marcus-what-could-possibly-go-wrong-with-1b452d-a1.

  45. Assessment publishedBy this siteGary Marcus

    It is not clear that anywhere near $7 trillion of capital is available for Altman's project.

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

    Assessment marcus-what-could-possibly-go-wrong-with-0c25c5-a1.

  46. Assessment publishedBy this siteGary Marcus

    If OpenAI gains direct access to user devices, the risk of it being hacked and the rewards to attackers will increase.

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

    Assessment marcus-what-could-possibly-go-wrong-with-09f7ef-a1.

  47. Assessment publishedBy this siteGary Marcus

    Without massive technical advances, chains of LLMs are unlikely to provide a stable and secure basis for agents.

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

    Assessment marcus-what-could-possibly-go-wrong-with-0849ec-a1.

  48. Assessment publishedBy this siteGary Marcus

    Altman will probably never raise the $7 trillion, even in ten yearly $700B tranches.

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

    Assessment marcus-what-could-possibly-go-wrong-with-065122-a1.

  49. Assessment publishedBy this siteGary Marcus

    Given sufficient data, deepfake tools can now fake anybody's appearance at any time for almost no cost.

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

    Assessment marcus-we-urgently-need-a-federal-law-forbidding-f67598-a1.

  50. Assessment publishedBy this siteGary Marcus

    OpenClaw can be used to phone people while pretending to be human, a new security risk.

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

    Assessment marcus-we-urgently-need-a-federal-law-forbidding-f324e2-a1.

  52. HeldBy this siteGary Marcus

    Held record marcus-we-urgently-need-a-federal-law-forbidding-c772df

    hedged language is not scored, even on a plain reading

  53. Assessment publishedBy this siteGary Marcus

    The US federal government is trying to undermine states' ability to pass and enforce AI regulation.

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

    Assessment marcus-we-urgently-need-a-federal-law-forbidding-ae82e3-a1.

  54. Assessment publishedBy this siteGary Marcus

    Scammers are among the first adopters of AI impersonation and deepfake tools and are already using them.

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

    Assessment marcus-we-urgently-need-a-federal-law-forbidding-a81599-a1.

  55. Assessment publishedBy this siteGary Marcus

    The US must pass federal laws forbidding machine output from being presented as human and develop the means to enforce them.

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

    Assessment marcus-we-urgently-need-a-federal-law-forbidding-6b97ae-a1.

  56. Assessment publishedBy this siteGary Marcus

    Corporate lobbyists must not be allowed to thwart efforts to regulate AI impersonation and deepfakes.

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

    Assessment marcus-we-urgently-need-a-federal-law-forbidding-4ca948-a1.

  57. Assessment publishedBy this siteGary Marcus

    A law forbidding the creation and passing along of AI 'counterfeit people' is now urgently needed.

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

    Assessment marcus-we-urgently-need-a-federal-law-forbidding-4bb483-a1.

  59. Assessment publishedBy this siteGary Marcus

    The current generation of AI agents cannot be trusted, as shown by incidents at both OpenAI and Anthropic.

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

    Assessment marcus-we-must-recall-open-ended-ai-agents-f1f80e-a1.

  60. HeldBy this siteGary Marcus

    Held record marcus-we-must-recall-open-ended-ai-agents-b085f0

    The anthropic screen reads the claim differently.

  61. Assessment publishedBy this siteGary Marcus

    Marcus's sense is that these AI companies, such as OpenAI and Anthropic, are wildly in over their heads.

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

    Assessment marcus-we-must-recall-open-ended-ai-agents-96cb9a-a1.

  62. Assessment publishedBy this siteGary Marcus

    Marcus has long felt that OpenAI is handling "this" (apparently the agent-caused incidents) inadequately.

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

    Assessment marcus-we-must-recall-open-ended-ai-agents-724ebe-a1.

  64. Assessment publishedBy this siteGary Marcus

    Each day the administration fails to take stronger action is a mistake, inviting worse problems.

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

    Assessment marcus-we-must-recall-open-ended-ai-agents-5c142b-a1.

  65. Assessment publishedBy this siteGary Marcus

    The administration should already have imposed a temporary recall on open-ended AI agents, which are obviously dangerous.

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

    Assessment marcus-we-must-recall-open-ended-ai-agents-351281-a1.

  66. Assessment publishedBy this siteGary Marcus

    Current AI agents should be removed from the market until they can be fixed.

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

    Assessment marcus-we-must-recall-open-ended-ai-agents-24b7a1-a1.

  67. HeldBy this siteGary Marcus

    Held record marcus-we-interrupt-this-programming-2bdec0

    The anthropic screen reads the claim differently.

  68. Assessment publishedBy this siteGary Marcus

    We waited too long to act against AI-generated misinformation and deepfakes in elections.

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

    Assessment marcus-we-cant-trust-the-fox-to-guard-the-fdcb5b-a1.

  70. Assessment publishedBy this siteGary Marcus

    AI needs independent oversight rather than self-regulation by the tech industry.

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

    Assessment marcus-we-cant-trust-the-fox-to-guard-the-bdccb1-a1.

  71. Assessment publishedBy this siteGary Marcus

    Because Silicon Valley is prone to hype, Washington needs independent scientists involved in AI policy discussions.

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

    Assessment marcus-we-cant-trust-the-fox-to-guard-the-b06f7e-a1.

  72. Assessment publishedBy this siteGary Marcus

    We need to act now to counter Silicon Valley's control.

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

    Assessment marcus-we-cant-trust-the-fox-to-guard-the-929b6b-a1.

  74. Assessment publishedBy this siteGary Marcus

    So far, Washington has not done enough as a counterweight to keep the captains of industry honest.

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

    Assessment marcus-we-cant-trust-the-fox-to-guard-the-749ccc-a1.

  75. Assessment publishedBy this siteGary Marcus

    If Schumer does not bring AI-related legislation to the Senate floor, we will be sitting ducks, meaning highly vulnerable.

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

    Assessment marcus-we-cant-trust-the-fox-to-guard-the-4a95dd-a1.

  77. Assessment publishedBy this siteGary Marcus

    If the White House treats tech CEOs like rock stars, those CEOs will determine AI policy.

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

    Assessment marcus-we-cant-trust-the-fox-to-guard-the-0b3d4f-a1.

  78. Assessment publishedBy this siteGary Marcus

    Without a coherent framework, US AI policy is headed toward poorly considered chaos.

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

    Assessment marcus-us-ai-policy-is-a-clumsy-mess-heres-f6d94e-a1.

  79. Assessment publishedBy this siteGary Marcus

    The US has 1,200 AI bills, roughly 150 of them enacted into law, but nothing that feels like a coherent AI policy.

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

    Assessment marcus-us-ai-policy-is-a-clumsy-mess-heres-d87591-a1.

  80. Assessment publishedBy this siteGary Marcus

    The situation of many US AI bills without anything that feels like a coherent AI policy is not good for anyone, businesses or consumers.

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

    Assessment marcus-us-ai-policy-is-a-clumsy-mess-heres-62edff-a1.

  82. HeldBy this siteGary Marcus

    Held record marcus-urgent-warning-black-mirror-has-entered-0ed891

    The final-text check failed: citations_correct.

  85. HeldBy this siteGary Marcus

    Held record marcus-unwoven-f9a60a-2

    The anthropic screen reads the claim differently.

  86. Assessment publishedBy this siteGary Marcus

    If CoreWeave's stock market debut goes badly, there may be shockwaves possibly affecting the whole AI industry.

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

    Assessment marcus-unwoven-656881-a1.

  87. Assessment publishedBy this siteGary Marcus

    If the CoreWeave IPO is a debacle, people may sour on NVidia too, and things could get wild quickly.

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

    Assessment marcus-unwoven-54f1ce-a1.

  88. Assessment publishedBy this siteGary Marcus

    Neurosymbolic AI might be a promising alternative to pure deep learning.

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

    Assessment marcus-two-years-later-deep-learning-is-fa70db-a1.

  89. HeldBy this siteGary Marcus

    Held record marcus-two-years-later-deep-learning-is-e3ce3f

    The anthropic screen reads the claim differently.

  90. Assessment publishedBy this siteGary Marcus

    As of March 2024, AI is still far from trustworthy.

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

    Assessment marcus-two-years-later-deep-learning-is-e0d223-a1.

  91. Assessment publishedBy this siteGary Marcus

    None of the points in the post mean that new innovations of some sort will not eventually arrive, or that AGI is impossible.

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

    Assessment marcus-two-years-later-deep-learning-is-cb37b9-a1.

  92. HeldBy this siteGary Marcus

    Held record marcus-two-years-later-deep-learning-is-bd0437

    The openai screen reads the claim differently.

  93. HeldBy this siteGary Marcus

    Held record marcus-two-years-later-deep-learning-is-bb65d6

    The anthropic screen reads the claim differently.

  94. Assessment publishedBy this siteGary Marcus

    Marcus still feels the AI field is mostly chasing ladders to the moon.

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

    Assessment marcus-two-years-later-deep-learning-is-b1b3c0-a1.

  95. HeldBy this siteGary Marcus

    Held record marcus-two-years-later-deep-learning-is-adbcc9

    The openai screen reads the claim differently.

  96. Assessment publishedBy this siteGary Marcus

    Getting better at next-word prediction through scaling does not amount to the deep comprehension true AI needs, and this is increasingly acknowledged.

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

    Assessment marcus-two-years-later-deep-learning-is-9e3bd0-a1.

  99. Assessment publishedBy this siteGary Marcus

    Scaling might not solve the fundamental problems of deep learning.

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

    Assessment marcus-two-years-later-deep-learning-is-5a22c7-a1.

  100. Assessment publishedBy this siteGary Marcus

    AI needs a paradigm shift beyond current deep learning to reach general intelligence.

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

    Assessment marcus-two-years-later-deep-learning-is-44e4a8-a1.

  101. Assessment publishedBy this siteGary Marcus

    As of March 2024, machines are still a long way from genuinely understanding human language.

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

    Assessment marcus-two-years-later-deep-learning-is-1860c1-a1.

  102. HeldBy this siteGary Marcus

    Held record marcus-two-years-later-deep-learning-is-031f1b

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

  103. Assessment publishedBy this siteGary Marcus

    By many accounts, AI-fueled scams have exploded, and there is no coherent US federal response to them.

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

    Assessment marcus-two-years-ago-today-in-ai-history-f5ce27-a1.

  104. Assessment publishedBy this siteGary Marcus

    Being beguiled by Sam Altman may turn out to have an enormous cost to humanity.

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

    Assessment marcus-two-years-ago-today-in-ai-history-de4e90-a1.

  105. Assessment publishedBy this siteGary Marcus

    The chance to avoid AI repeating social media's regulatory failure, where big tech got its way at society's expense, is gone.

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

    Assessment marcus-two-years-ago-today-in-ai-history-d82d88-a1.

  107. Assessment publishedBy this siteGary Marcus

    While Altman publicly supported AI regulation in 2023, OpenAI lobbied to water down the EU AI Act.

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

    Assessment marcus-two-years-ago-today-in-ai-history-9cfc7b-a1.

  108. Assessment publishedBy this siteGary Marcus

    US government officials now downplay rather than voice concern about AI risks.

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

    Assessment marcus-two-years-ago-today-in-ai-history-9ad0f6-a1.

  109. HeldBy this siteGary Marcus

    Held record marcus-two-years-ago-today-in-ai-history-8f961d

    The anthropic screen reads the claim differently.

  110. HeldBy this siteGary Marcus

    Held record marcus-two-years-ago-today-in-ai-history-76829e

    The anthropic screen reads the claim differently.

  111. Assessment publishedBy this siteGary Marcus

    Since 2023 OpenAI has pushed for free training materials and exemption from copyright law, contrary to Altman's stated support for content creators.

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

    Assessment marcus-two-years-ago-today-in-ai-history-74dd27-a1.

  112. Assessment publishedBy this siteGary Marcus

    Congress has passed no meaningful AI regulation and is trying to prevent states from passing meaningful AI regulation.

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

    Assessment marcus-two-years-ago-today-in-ai-history-5f86d2-a2.

  113. Assessment publishedBy this siteGary Marcus

    AI-fueled discrimination in job decisions is likely far more widespread in 2025 than in 2023.

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

    Assessment marcus-two-years-ago-today-in-ai-history-5e47f1-a1.

  114. HeldBy this siteGary Marcus

    Held record marcus-two-years-ago-today-in-ai-history-4f68cb

    The final-text check failed: citations_correct.

  115. HeldBy this siteGary Marcus

    Held record marcus-two-years-ago-today-in-ai-history-3f2e45

    The openai screen doubts the attribution.

  116. HeldBy this siteGary Marcus

    Held record marcus-two-years-ago-today-in-ai-history-384068

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

  117. Assessment publishedBy this siteGary Marcus

    Washington no longer seriously considers acting on AI risks.

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

    Assessment marcus-two-years-ago-today-in-ai-history-3776fa-a1.

  118. Assessment publishedBy this siteGary Marcus

    Sam Altman is doing everything he can to stop meaningful AI regulation.

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

    Assessment marcus-two-years-ago-today-in-ai-history-2e3897-a2.

  119. Assessment publishedBy this siteGary Marcus

    US government AI discourse has replaced regulation with 'innovation', meaning helping companies regardless of cost to citizens.

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

    Assessment marcus-two-years-ago-today-in-ai-history-1ec279-a1.

  120. Assessment publishedBy this siteGary Marcus

    Nothing is being done about political or medical AI-generated misinformation.

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

    Assessment marcus-two-years-ago-today-in-ai-history-0cc6cf-a1.

  121. HeldBy this siteGary Marcus

    Held record marcus-two-years-ago-today-in-ai-history-0c1c79

    The anthropic review does not accept e1 as supports.

  122. Assessment publishedBy this siteGary Marcus

    It is not just United: one of the biggest applications of AI seems to be ripping off customers.

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

    Assessment marcus-two-visions-of-ais-future-cbc9f2-a1.

  123. HeldBy this siteGary Marcus

    Held record marcus-two-visions-of-ais-future-7f3ec8

    The openai screen reads the claim differently.

  124. Assessment publishedBy this siteGary Marcus

    If AI is left unregulated, it will probably lead to dystopia.

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

    Assessment marcus-two-visions-of-ais-future-4dc30c-a1.

  125. Assessment publishedBy this siteGary Marcus

    The US could easily end up with too many data centers and chips that rapidly lose value.

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

    Assessment marcus-two-takes-on-ai-and-the-future-of-eaad4d-a1.

  126. Assessment publishedBy this siteGary Marcus

    Many others have found that LLMs are not quite as good as advertised.

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

    Assessment marcus-two-takes-on-ai-and-the-future-of-bc568e-a1.

  127. HeldBy this siteGary Marcus

    Held record marcus-two-takes-on-ai-and-the-future-of-b3281a

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

  128. Assessment publishedBy this siteGary Marcus

    Going all in on generative AI without intellectual and economic diversification is likely an immense mistake for the nation.

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

    Assessment marcus-two-takes-on-ai-and-the-future-of-aef803-a1.

  129. Assessment publishedBy this siteGary Marcus

    The July METR study showed that AI coding tools sometimes impair productivity rather than increase it.

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

    Assessment marcus-two-takes-on-ai-and-the-future-of-a11d10-a1.

  130. Assessment publishedBy this siteGary Marcus

    In the worst case, there may be 'too big to fail' claims and calls for government bailouts of the AI industry.

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

    Assessment marcus-two-takes-on-ai-and-the-future-of-855cd6-a1.

  132. HeldBy this siteGary Marcus

    Held record marcus-two-takes-on-ai-and-the-future-of-6e7e4f

    e1 is dated 2025-12, which is not inside the period described (2025-10-01 to 2025-12-23); it can only be context

  133. Assessment publishedBy this siteGary Marcus

    Speculative US investment in chips and data centers conflicts with growing disappointment in generative AI.

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

    Assessment marcus-two-takes-on-ai-and-the-future-of-52f116-a1.

  134. Assessment publishedBy this siteGary Marcus

    Trouble is brewing even in coding, the flagship application of generative AI.

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

    Assessment marcus-two-takes-on-ai-and-the-future-of-365d7f-a1.

  136. HeldBy this siteGary Marcus

    Held record marcus-two-nobel-prizes-for-ai-and-two-paths-b73323

    The openai review reads the claim differently.

  137. Assessment publishedBy this siteGary Marcus

    People will probably keep questioning Hinton's Nobel award for a long time.

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

    Assessment marcus-two-nobel-prizes-for-ai-and-two-paths-83011b-a1.

  139. Assessment publishedBy this siteGary Marcus

    Hassabis's open-mindedness will serve the AI field well.

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

    Assessment marcus-two-nobel-prizes-for-ai-and-two-paths-63e828-a1.

  140. HeldBy this siteGary Marcus

    Held record marcus-two-nobel-prizes-for-ai-and-two-paths-5c4b69

    e4 is dated 2024, which is not inside the period described (2024-01-01 to 2024-10-09); it can only be context; e5 is dated 2024, which is not inside the period described (2024-01-01 to 2024-10-09); it can only be context

  141. Assessment publishedBy this siteGary Marcus

    The AI field should move to a new approach beyond pure end-to-end neural networks, as Hassabis recognises.

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

    Assessment marcus-two-nobel-prizes-for-ai-and-two-paths-5805a9-a1.

  142. HeldBy this siteGary Marcus

    Held record marcus-two-nobel-prizes-for-ai-and-two-paths-579505

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

  143. HeldBy this siteGary Marcus

    Held record marcus-two-nobel-prizes-for-ai-and-two-paths-27d412

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

  145. HeldBy this siteGary Marcus

    Held record marcus-two-nobel-prizes-for-ai-and-two-paths-0f5180

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

  146. Assessment publishedBy this siteGary Marcus

    Unlike Hinton's prize, the chemistry Nobel won by Hassabis and shared with John Jumper is a slam dunk, meaning clearly deserved.

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

    Assessment marcus-two-nobel-prizes-for-ai-and-two-paths-085170-a1.

  147. Assessment publishedBy this siteGary Marcus

    If AI companies set the rules on their own, the needed safeguards are unlikely to result.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-f66017-a1.

  148. Assessment publishedBy this siteGary Marcus

    The AI future depends largely on government standing up to big tech, which depends on independent voices being at the table.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-f3153d-a1.

  149. Assessment publishedBy this siteGary Marcus

    With the right AI regulation things could go well; with the wrong regulation things could go badly.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-ee5655-a1.

  150. Assessment publishedBy this siteGary Marcus

    AI needs enforceable regulation rather than photo opportunities.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-e806c2-a1.

  151. Assessment publishedBy this siteGary Marcus

    If big tech writes AI rules without outside input, the right rules are unlikely to result.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-e3ae19-a1.

  152. HeldBy this siteGary Marcus

    Held record marcus-two-models-of-ai-oversight-and-how-ced689

    The openai screen reads the claim differently.

  153. Assessment publishedBy this siteGary Marcus

    Government should require transparency and safety from AI companies and stop them from flouting those requirements.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-ce2f2a-a1.

  154. Assessment publishedBy this siteGary Marcus

    Governments should require AI systems to use copyrighted content only from opt-in providers and to label all machine-generated content.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-c91983-a1.

  158. Assessment publishedBy this siteGary Marcus

    Government photo opportunities with tech moguls without independent voices send the wrong message.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-81a2bd-a1.

  160. Assessment publishedBy this siteGary Marcus

    It is still possible to build a very positive AI future.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-74859f-a1.

  161. Assessment publishedBy this siteGary Marcus

    The AI industry should not be left to regulate itself, just as pharma is not.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-6f0834-a1.

  162. Assessment publishedBy this siteGary Marcus

    There is a risk that governments make AI rules that entrench incumbents while doing too little for humanity.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-4a5ef3-a1.

  163. HeldBy this siteGary Marcus

    Held record marcus-two-models-of-ai-oversight-and-how-1e2674

    e1 is dated 2023-01, which is not inside the period described (2023-01-23 to 2023-06-08); it can only be context

  165. Assessment publishedBy this siteGary Marcus

    Bing has defamed people, and in doing so has misread articles as saying the opposite of what they actually say.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-16523e-a1.

  166. Assessment publishedBy this siteGary Marcus

    What really matters is the regulation that governments around the world come up with.

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

    Assessment marcus-two-models-of-ai-oversight-and-how-025d48-a1.

  167. Assessment publishedBy this siteGary Marcus

    Being long-term optimistic about technology does not mean one has to ignore the short-term or long-term risks of technology.

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

    Assessment marcus-two-kinds-of-techno-optimism-9e224d-a1.

  169. Assessment publishedBy this siteGary Marcus

    A positive technological future has to be actively worked for, not just declared inevitable while hoping for the best.

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

    Assessment marcus-two-kinds-of-techno-optimism-116ed6-a1.

  170. Assessment publishedBy this siteGary Marcus

    Tools for spreading automatically generated misinformation are becoming more efficient.

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

    Assessment marcus-two-frightening-things-before-breakfast-edd8da-a2.

  171. Assessment publishedBy this siteGary Marcus

    There is almost no regulation of AI chatbots.

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

    Assessment marcus-two-frightening-things-before-breakfast-c13aae-a1.

  172. Assessment publishedBy this siteGary Marcus

    Microsoft is, roughly, free to do whatever it wants with Bing's AI.

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

    Assessment marcus-two-frightening-things-before-breakfast-a82400-a1.

  173. Assessment publishedBy this siteGary Marcus

    People should be worried about the dangers of chatbots right now, not at some future date.

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

    Assessment marcus-two-frightening-things-before-breakfast-96f96d-a1.

  174. Assessment publishedBy this siteGary Marcus

    An unsophisticated user might be taken in by the Bing chatbot and harmed if the bot proves fickle.

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

    Assessment marcus-two-frightening-things-before-breakfast-8bc125-a1.

  175. HeldBy this siteGary Marcus

    Held record marcus-two-frightening-things-before-breakfast-590416

    The openai screen doubts the attribution.

  176. Assessment publishedBy this siteGary Marcus

    Retraining GPT-3 to take on a new political slant costs far less than half a million dollars.

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

    Assessment marcus-two-frightening-things-before-breakfast-57e353-a2.

  177. HeldBy this siteGary Marcus

    Held record marcus-two-frightening-things-before-breakfast-569c6c

    The anthropic review does not accept e1 as supports.

  178. Assessment publishedBy this siteGary Marcus

    Bing's chatbot falsely claimed that Marcus believes Bing's AI is more accurate and reliable than Bard.

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

    Assessment marcus-two-frightening-things-before-breakfast-5252b4-a1.

  179. HeldBy this siteGary Marcus

    Held record marcus-two-frightening-things-before-breakfast-2e0b22

    The openai screen reads the claim differently.

  180. Assessment publishedBy this siteGary Marcus

    ChatGPT has no political ideas, so strictly speaking it does not literally lean left or right.

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

    Assessment marcus-two-frightening-things-before-breakfast-2a59eb-a1.

  181. HeldBy this siteGary Marcus

    Held record marcus-two-frightening-things-before-breakfast-0e23b0

    The openai review reads the claim differently.

  182. Assessment publishedBy this siteGary Marcus

    OpenAI's Astra may not be the breakthrough OpenAI portrays it as.

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

    Assessment marcus-two-critical-updates-re-astra-and-f50e4b-a1.

  183. HeldBy this siteGary Marcus

    Held record marcus-two-critical-updates-re-astra-and-f1f1cc

    The anthropic screen reads the claim differently.

  184. Assessment publishedBy this siteGary Marcus

    Marcus is often left with the feeling that OpenAI's public-facing communications are better marketing than they are science.

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

    Assessment marcus-two-critical-updates-re-astra-and-e80903-a1.

  185. HeldBy this siteGary Marcus

    Held record marcus-two-critical-updates-re-astra-and-d92610

    The anthropic screen reads the claim differently.

  186. Assessment publishedBy this siteGary Marcus

    How good Astra is on open-ended real-world problems very much remains to be seen.

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

    Assessment marcus-two-critical-updates-re-astra-and-8e1a1b-a2.

  187. Assessment publishedBy this siteGary Marcus

    OpenAI faces huge economic and competitive pressures and needs a win.

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

    Assessment marcus-two-critical-updates-re-astra-and-8cae47-a1.

  188. Assessment publishedBy this siteGary Marcus

    OpenAI did not report which problems Astra tried and failed at, giving a numerator without a denominator.

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

    Assessment marcus-two-critical-updates-re-astra-and-78af93-a2.

  189. HeldBy this siteGary Marcus

    Held record marcus-two-critical-updates-re-astra-and-4d4025

    The anthropic screen reads the claim differently.

  190. Assessment publishedBy this siteGary Marcus

    Astra is an advance but may well turn out to be merely incremental relative to other recent models.

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

    Assessment marcus-two-critical-updates-re-astra-and-332516-a1.

  191. HeldBy this siteGary Marcus

    Held record marcus-two-critical-updates-re-astra-and-2cf9eb

    The anthropic review reads the claim differently.

  193. HeldBy this siteGary Marcus

    Held record marcus-two-critical-updates-re-astra-and-14ba8c

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

  194. Assessment publishedBy this siteGary Marcus

    Astra's math results may have stemmed mainly from a (probably human) idea to run a particular search rather than from innovation in Astra itself.

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

    Assessment marcus-two-critical-updates-re-astra-and-126fa4-a1.

  195. HeldBy this siteGary Marcus

    Held record marcus-two-critical-updates-re-astra-and-00a6ae

    The anthropic review does not accept e1 as supports.

  196. HeldBy this siteGary Marcus

    Held record marcus-turns-out-generative-ai-was-a-scam-fdf6b2

    The final-text check failed: citations_correct.

  197. HeldBy this siteGary Marcus

    Held record marcus-turns-out-generative-ai-was-a-scam-eb2a3c

    e1 is dated 2026-02, which is not inside the period described (2026-02-01 to 2026-02-23); it can only be context; e2 is dated 2026-02, which is not inside the period described (2026-02-01 to 2026-02-23); it can only be context

  198. HeldBy this siteGary Marcus

    Held record marcus-turns-out-generative-ai-was-a-scam-dde00e

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

  199. HeldBy this siteGary Marcus

    Held record marcus-turns-out-generative-ai-was-a-scam-955997

    The anthropic review does not accept e1 as supports.

  200. Assessment publishedBy this siteGary Marcus

    Large language models often produce low-quality 'workslop'.

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

    Assessment marcus-turns-out-generative-ai-was-a-scam-8d6cdb-a1.