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

Changes, page 11 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 11 of 37: changes 2001 to 2200 of 7383, newest first. Dates are when the change was recorded.

  1. Assessment publishedBy this siteGary Marcus

    GM's suspension of Cruise is likely to have a major impact on the self-driving industry.

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

    Assessment marcus-why-driverless-cars-mightor-might-41b3a5-a1.

  2. Assessment publishedBy this siteGary Marcus

    Generative search and personalized agents probably also need much smarter, reasoning, non-hallucinating AI that comes from fundamental research.

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

    Assessment marcus-why-driverless-cars-mightor-might-411585-a1.

  3. Assessment publishedBy this siteGary Marcus

    The sooner the field recognizes that there are options beyond current AI approaches, the faster it will reach reliable self-driving, search and agents.

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

    Assessment marcus-why-driverless-cars-mightor-might-37dc5e-a1.

  4. HeldBy this siteGary Marcus

    Held record marcus-why-driverless-cars-mightor-might-34953f

    The final-text check failed: proposition_faithful.

  5. Assessment publishedBy this siteGary Marcus

    LLMs have such a superficial understanding of their own output that they can't even begin to fact-check it.

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

    Assessment marcus-why-do-large-language-models-hallucinate-fc26ca-a1.

  6. Assessment publishedBy this siteGary Marcus

    The field should not put all its eggs in the LLM basket; doing so is delusional.

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

    Assessment marcus-why-do-large-language-models-hallucinate-f1b4a6-a1.

  7. Assessment publishedBy this siteGary Marcus

    AI will eventually get to something better than LLMs.

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

    Assessment marcus-why-do-large-language-models-hallucinate-f1b4a6-2-a1.

  9. HeldBy this siteGary Marcus

    Held record marcus-why-do-large-language-models-hallucinate-edbb3e

    e1 is dated 2025, which is not inside the period described (2025-01-01 to 2025-05-05); it can only be context

  11. Assessment publishedBy this siteGary Marcus

    After 8 years and roughly half a trillion dollars, LLMs still lack a grasp of reality.

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

    Assessment marcus-why-do-large-language-models-hallucinate-6006ff-a1.

  12. Assessment publishedBy this siteGary Marcus

    LLMs on their own may never be able to reliably check their output against basic sources.

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

    Assessment marcus-why-do-large-language-models-hallucinate-5551a6-a1.

  13. Assessment publishedBy this siteGary Marcus

    LLMs are useful e.g. for coding autocomplete, but their human-sounding outputs do not imply human-like computation.

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

    Assessment marcus-why-do-large-language-models-hallucinate-52167d-a1.

  14. Assessment publishedBy this siteGary Marcus

    Even the best performance in the cited study, a 15% hallucination rate, is pathetic for an open-book exam where the sources were supplied.

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

    Assessment marcus-why-do-large-language-models-hallucinate-18057a-a1.

  15. Assessment publishedBy this siteGary Marcus

    The US-China generative AI race will far more likely end in a tie than in either side dominating.

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

    Assessment marcus-why-chatgpt-cant-be-trusted-with-bda53b-a1.

  17. Assessment publishedBy this siteGary Marcus

    The idea of using LLMs to plot strategy in rapidly-changing environments like war is laughable.

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

    Assessment marcus-why-chatgpt-cant-be-trusted-with-21bf3e-a1.

  18. Assessment publishedBy this siteGary Marcus

    Businesses should not trust LLMs, in large part because LLMs are unreliable in the face of novelty.

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

    Assessment marcus-why-chatgpt-cant-be-trusted-with-1d9a76-a1.

  19. HeldBy this siteGary Marcus

    Held record marcus-why-chatgpt-cant-be-trusted-with-1ad410

    The openai screen reads the claim differently.

  20. Assessment publishedBy this siteGary Marcus

    SB-1047 must pass the California Assembly and be signed by Governor Newsom rather than vetoed.

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

    Assessment marcus-why-californias-ai-safety-bill-should-e98313-a1.

  21. Assessment publishedBy this siteGary Marcus

    SB-1047's main value may be deterring irresponsible AI development by clarifying liability.

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

    Assessment marcus-why-californias-ai-safety-bill-should-e2791c-a1.

  22. HeldBy this siteGary Marcus

    Held record marcus-why-californias-ai-safety-bill-should-e190f4

    The anthropic screen doubts the attribution.

  24. Assessment publishedBy this siteGary Marcus

    SB-1047 deserves support even in its weakened form.

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

    Assessment marcus-why-californias-ai-safety-bill-should-b01259-a1.

  25. Assessment publishedBy this siteGary Marcus

    AI companies may gamble on risky behavior and let lawyers handle consequences afterward despite SB-1047.

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

    Assessment marcus-why-californias-ai-safety-bill-should-8606d2-a1.

  28. Assessment publishedBy this siteGary Marcus

    Most or all major big tech companies joined a lobbying group that fought SB-1047 hard despite broad public support for it.

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

    Assessment marcus-why-californias-ai-safety-bill-should-619c29-a2.

  29. Assessment publishedBy this siteGary Marcus

    In Marcus's view, many of the arguments made against SB-1047 by Andreessen Horowitz and others were based on misrepresentation.

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

    Assessment marcus-why-californias-ai-safety-bill-should-5b7135-a1.

  30. Assessment publishedBy this siteGary Marcus

    A messy state-by-state patchwork of AI laws could stifle AI development.

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

    Assessment marcus-why-californias-ai-safety-bill-should-4cef82-a1.

  31. Assessment publishedBy this siteGary Marcus

    SB-1047 cannot be the last AI bill in California or the US if Americans are to be protected from AI's many risks.

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

    Assessment marcus-why-californias-ai-safety-bill-should-4683d5-a1.

  33. Assessment publishedBy this siteGary Marcus

    The US needs a comprehensive, federal approach to AI regulation; SB-1047 is just a start.

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

    Assessment marcus-why-californias-ai-safety-bill-should-37a373-a1.

  34. Assessment publishedBy this siteGary Marcus

    Very large companies may overwhelm SB-1047's reasonable care standard, and legal actions under it may take many years.

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

    Assessment marcus-why-californias-ai-safety-bill-should-307af4-a1.

  35. Assessment publishedBy this siteGary Marcus

    SB-1047 provides important whistleblower protections, which, as many ex-OpenAI employees have made clear, are critical.

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

    Assessment marcus-why-californias-ai-safety-bill-should-2c5399-a1.

  36. Assessment publishedBy this siteGary Marcus

    OpenAI talked big about investing in safety research and then failed to deliver on those promises.

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

    Assessment marcus-why-californias-ai-safety-bill-should-270f2b-a2.

  39. HeldBy this siteGary Marcus

    Held record marcus-which-ceo-will-be-the-last-to-see-869232

    The openai screen reads the claim differently.

  40. HeldBy this siteGary Marcus

    Held record marcus-which-ceo-will-be-the-last-to-see-8455b7

    The anthropic screen doubts the attribution.

  41. Assessment publishedBy this siteGary Marcus

    Satya Nadella has publicly acknowledged diminishing returns.

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

    Assessment marcus-which-ceo-will-be-the-last-to-see-6749d2-a1.

  43. HeldBy this siteGary Marcus

    Held record marcus-wheres-waldo-the-elephant-in-the-f287ef-2

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

  44. Assessment publishedBy this siteGary Marcus

    Training on terabytes of images enables impressive image generation but does not guarantee the system understands how images relate to the world.

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

    Assessment marcus-wheres-waldo-the-elephant-in-the-f10eb1-a1.

  45. HeldBy this siteGary Marcus

    Held record marcus-wheres-waldo-the-elephant-in-the-ea5480

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

  46. Assessment publishedBy this siteGary Marcus

    Efforts to use generative AI systems to address copyright issues will run into problems of discomprehension.

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

    Assessment marcus-wheres-waldo-the-elephant-in-the-d71263-a1.

  47. Assessment publishedBy this siteGary Marcus

    The apparent quality of LLMs is only skin deep.

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

    Assessment marcus-wheres-waldo-the-elephant-in-the-d4a8a6-a1.

  48. HeldBy this siteGary Marcus

    Held record marcus-wheres-waldo-the-elephant-in-the-ba67d3

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

  50. Assessment publishedBy this siteGary Marcus

    AI errors of discomprehension may soon become more worrisome as AI is used in war.

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

    Assessment marcus-wheres-waldo-the-elephant-in-the-6a3e4f-a1.

  51. Assessment publishedBy this siteGary Marcus

    Generative AI used for source attribution will sometimes work but will probably sometimes produce outrageous errors.

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

    Assessment marcus-wheres-waldo-the-elephant-in-the-5f23d0-a1.

  52. Assessment publishedBy this siteGary Marcus

    Image generators still do not really understand how the world works.

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

    Assessment marcus-wheres-waldo-the-elephant-in-the-50934d-a1.

  53. Assessment publishedBy this siteGary Marcus

    Errors of discomprehension in neural-network-based systems will not disappear any time soon.

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

    Assessment marcus-wheres-waldo-the-elephant-in-the-3e5dae-a1.

  56. HeldBy this siteGary Marcus

    Held record marcus-where-we-are-right-now-on-open-source-c2e824

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

  57. Assessment publishedBy this siteGary Marcus

    It is terrifying that a single company can unilaterally make the open source AI decision for all of humanity.

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

    Assessment marcus-where-we-are-right-now-on-open-source-bcc56a-a1.

  58. Assessment publishedBy this siteGary Marcus

    There is a nonzero chance that the company's open source AI push has put everyone at massive risk without oversight.

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

    Assessment marcus-where-we-are-right-now-on-open-source-990f69-a1.

  59. Assessment publishedBy this siteGary Marcus

    We ought to be a bit more patient about comparing the risks and benefits of open source AI.

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

    Assessment marcus-where-we-are-right-now-on-open-source-797d72-a1.

  60. HeldBy this siteGary Marcus

    Held record marcus-where-we-are-right-now-on-open-source-314303

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

  61. Assessment publishedBy this siteGary Marcus

    No one knows for certain that open source AI has no serious possible consequences, such as for bioweapons or misinformation generation.

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

    Assessment marcus-where-we-are-right-now-on-open-source-1c7741-a2.

  62. HeldBy this siteGary Marcus

    Held record marcus-where-is-ai-headed-8-perspectives-f4728b

    the horizon on 2026-10-28 is open_horizon, not not_due

  63. HeldBy this siteGary Marcus

    Held record marcus-where-is-ai-headed-8-perspectives-a008fb

    the conclusive event is after the evidence cutoff

  64. Assessment publishedBy this siteGary Marcus

    Generative AI investment of about $50B against $3B in revenue is obviously not sustainable.

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

    Assessment marcus-when-will-the-genai-bubble-burst-ef2606-a2.

  65. Assessment publishedBy this siteGary Marcus

    Because everyone is building similar models, generative AI companies have no moat, leading to heavy price competition.

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

    Assessment marcus-when-will-the-genai-bubble-burst-e999b2-a1.

  66. HeldBy this siteGary Marcus

    Held record marcus-when-will-the-genai-bubble-burst-e8b82b

    e1 is decisive for a financial claim but states no metric scope; e2 is decisive for a financial claim but states no metric scope

  67. Assessment publishedBy this siteGary Marcus

    OpenAI is unlikely to release a mindblowing GPT model in 2024.

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

    Assessment marcus-when-will-the-genai-bubble-burst-e843ae-a1.

  68. Assessment publishedBy this siteGary Marcus

    GPT-5 may be more satisfying but is likely even more expensive to train and operate than current models.

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

    Assessment marcus-when-will-the-genai-bubble-burst-e53196-a1.

  69. HeldBy this siteGary Marcus

    Held record marcus-when-will-the-genai-bubble-burst-cba196

    hedged language is not scored, even on a plain reading

  70. Assessment publishedBy this siteGary Marcus

    Even with a mindblowing GPT-5 in 2024, GenAI might implode later once people realize there is no killer app to justify rising costs.

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

    Assessment marcus-when-will-the-genai-bubble-burst-a77173-a1.

  71. Assessment publishedBy this siteGary Marcus

    Core generative AI problems like hallucinations are not going away.

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

    Assessment marcus-when-will-the-genai-bubble-burst-99c8c9-a1.

  72. HeldBy this siteGary Marcus

    Held record marcus-when-will-the-genai-bubble-burst-82724e

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

  73. HeldBy this siteGary Marcus

    Held record marcus-when-will-the-genai-bubble-burst-6c9808

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

  74. Assessment publishedBy this siteGary Marcus

    Marcus increasingly believes his August 2023 call that generative AI may turn out to be a dud was correct, though he is not yet claiming victory.

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

    Assessment marcus-when-will-the-genai-bubble-burst-5994ac-a1.

  75. Assessment publishedBy this siteGary Marcus

    Coding assistance is probably the closest thing to a generative AI killer app, but its income is not large enough to cover chip, legal and other costs.

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

    Assessment marcus-when-will-the-genai-bubble-burst-57b445-a1.

  76. Assessment publishedBy this siteGary Marcus

    Hype around generative AI is a risk factor: the more hype, the bigger the fall if expectations are not met.

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

    Assessment marcus-when-will-the-genai-bubble-burst-5633fe-a1.

  77. Assessment publishedBy this siteGary Marcus

    Hallucinations are inherent to how LLMs make predictions and to their inability to look things up and reason over them.

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

    Assessment marcus-when-will-the-genai-bubble-burst-457817-a1.

  78. Assessment publishedBy this siteGary Marcus

    Most uses of LLMs so far have been for testing rather than production systems.

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

    Assessment marcus-when-will-the-genai-bubble-burst-4518bf-a1.

  79. Assessment publishedBy this siteGary Marcus

    The generative AI industry is based on hype and the hope that persistent LLM problems are about to be solved.

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

    Assessment marcus-when-will-the-genai-bubble-burst-2f1fdb-a1.

  80. Assessment publishedBy this siteGary Marcus

    Bigger models have higher costs, and current-sized models are already expensive and not reliable enough for many jobs.

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

    Assessment marcus-when-will-the-genai-bubble-burst-2b9be3-a1.

  81. Assessment publishedBy this siteGary Marcus

    GenAI will persist and find some use, but valuations and excitement may dissipate until genuine advances are made.

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

    Assessment marcus-when-will-the-genai-bubble-burst-299d4d-a1.

  82. Assessment publishedBy this siteGary Marcus

    Generative AI as currently envisioned may never come together.

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

    Assessment marcus-when-will-the-genai-bubble-burst-27d69f-a1.

  83. Assessment publishedBy this siteGary Marcus

    Reid Hoffman's optimistic September 2023 AI predictions had not come true as of March 2024.

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

    Assessment marcus-when-will-the-genai-bubble-burst-27395f-a2.

  85. Assessment publishedBy this siteGary Marcus

    If OpenAI releases a truly mindblowing GPT-5 in 2024, hype will increase and the GenAI bubble will not burst soon.

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

    Assessment marcus-when-will-the-genai-bubble-burst-13612e-a1.

  87. HeldBy this siteGary Marcus

    Held record marcus-when-looked-at-carefully-openais-dccd07

    The final-text check failed: citations_correct.

  90. HeldBy this siteGary Marcus

    Held record marcus-when-looked-at-carefully-openais-bc42a6

    The final-text check failed: citations_correct.

  92. HeldBy this siteGary Marcus

    Held record marcus-when-looked-at-carefully-openais-abc924

    The anthropic screen reads the claim differently.

  93. Assessment publishedBy this siteGary Marcus

    It is currently unknown whether open-source models without guardrails are more useful to bad actors than commercial models with guardrails.

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

    Assessment marcus-when-looked-at-carefully-openais-ab43ec-a1.

  94. Assessment publishedBy this siteGary Marcus

    Because the paper has an obvious a priori hypothesis, a Bonferroni correction is unnecessary and should not have been used.

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

    Assessment marcus-when-looked-at-carefully-openais-993a54-a1.

  95. Assessment publishedBy this siteGary Marcus

    There is no reason to take much comfort from OpenAI's threshold analysis.

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

    Assessment marcus-when-looked-at-carefully-openais-962c11-a1.

  96. Assessment publishedBy this siteGary Marcus

    Readers, including OpenAI, ought to be pretty worried by the accuracy results in the paper.

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

    Assessment marcus-when-looked-at-carefully-openais-8771bc-a1.

  99. HeldBy this siteGary Marcus

    Held record marcus-when-looked-at-carefully-openais-7624ca

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

  100. Assessment publishedBy this siteGary Marcus

    AI risk-benefit analyses need to be based on studies that are both methodologically sound and accurately interpreted.

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

    Assessment marcus-when-looked-at-carefully-openais-74b900-a1.

  102. HeldBy this siteGary Marcus

    Held record marcus-when-looked-at-carefully-openais-7298d7

    The anthropic review reads the claim differently.

  103. Assessment publishedBy this siteGary Marcus

    With a bigger sample size, the author doubts there would be any question that the effect of LLM access is statistically significant.

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

    Assessment marcus-when-looked-at-carefully-openais-5f0890-a1.

  106. Assessment publishedBy this siteGary Marcus

    In OpenAI's study, participants with both internet and LLM access were more accurate than those without in virtually every condition.

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

    Assessment marcus-when-looked-at-carefully-openais-596625-a1.

  107. Assessment publishedBy this siteGary Marcus

    Footnote C of OpenAI's paper, which says the results are not statistically significant, is extremely dubious.

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

    Assessment marcus-when-looked-at-carefully-openais-53e8d0-a1.

  109. HeldBy this siteGary Marcus

    Held record marcus-when-looked-at-carefully-openais-4524b4

    The anthropic screen reads the claim differently.

  110. Assessment publishedBy this siteGary Marcus

    In the bioweapons domain, even a tiny effect of LLMs on capability could be world-changing in a deeply negative way.

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

    Assessment marcus-when-looked-at-carefully-openais-179b2a-a1.

  111. Assessment publishedBy this siteGary Marcus

    The media and public must recognize that company white papers are not peer-reviewed, and peer review is essential for safety-critical questions.

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

    Assessment marcus-when-looked-at-carefully-openais-12b81c-a1.

  113. HeldBy this siteGary Marcus

    Held record marcus-when-looked-at-carefully-openais-04e71c

    The final-text check failed: proposition_faithful.

  115. Assessment publishedBy this siteGary Marcus

    ChatGPT may come to be seen as AI's greatest publicity stunt, and its ethical guardrails will take years to perfect.

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

    Assessment marcus-when-it-comes-to-security-llms-are-a95267-a1.

  116. Assessment publishedBy this siteGary Marcus

    LLM chatbots open wide possibilities for propaganda, troll farms and fake websites that erode trust online.

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

    Assessment marcus-when-it-comes-to-security-llms-are-7d21ad-a1.

  117. Assessment publishedBy this siteGary Marcus

    The state of LLM chatbot safety is a disaster in the making.

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

    Assessment marcus-when-it-comes-to-security-llms-are-64cfaf-a1.

  119. Assessment publishedBy this siteGary Marcus

    The proposed 10-year federal preemption of state AI regulation is likely at odds with the 10th Amendment.

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

    Assessment marcus-when-it-comes-to-ai-policy-congress-f4a3e5-a1.

  120. HeldBy this siteGary Marcus

    Held record marcus-when-it-comes-to-ai-policy-congress-e7e05c

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

  121. HeldBy this siteGary Marcus

    Held record marcus-when-it-comes-to-ai-policy-congress-e7e05c-2

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

  122. Assessment publishedBy this siteGary Marcus

    Federal preemption of state AI law risks a dangerous precedent where the party with federal power overrides states' 10th Amendment powers.

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

    Assessment marcus-when-it-comes-to-ai-policy-congress-e74a44-a1.

  124. Assessment publishedBy this siteGary Marcus

    Washington is unlikely to pass a comprehensive privacy or AI law with teeth.

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

    Assessment marcus-when-it-comes-to-ai-policy-congress-e022e3-a1.

  125. Assessment publishedBy this siteGary Marcus

    AI leaderboards are no longer dominated by any single country.

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

    Assessment marcus-when-it-comes-to-ai-policy-congress-dfdbf3-a1.

  126. HeldBy this siteGary Marcus

    Held record marcus-when-it-comes-to-ai-policy-congress-d913de

    The anthropic review reads the claim differently.

  127. Assessment publishedBy this siteGary Marcus

    A decade of AI deregulation isn't a path forward; it is an abdication of responsibility.

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

    Assessment marcus-when-it-comes-to-ai-policy-congress-9ea00a-a1.

  128. HeldBy this siteGary Marcus

    Held record marcus-when-it-comes-to-ai-policy-congress-84464d

    e3 is dated 2025-05, which is not inside the period described (2024-01-01 to 2025-05-14); it can only be context

  129. HeldBy this siteGary Marcus

    Held record marcus-when-it-comes-to-ai-policy-congress-7fc11e

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

  130. HeldBy this siteGary Marcus

    Held record marcus-when-it-comes-to-ai-policy-congress-7ad22c

    The anthropic screen reads the claim differently.

  131. Assessment publishedBy this siteGary Marcus

    Americans who support AI oversight should contact their members of Congress now to oppose the preemption provision.

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

    Assessment marcus-when-it-comes-to-ai-policy-congress-7ad22c-2-a1.

  132. Assessment publishedBy this siteGary Marcus

    Silicon Valley loves to invoke China to get better terms, but this probably won't make any difference.

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

    Assessment marcus-when-it-comes-to-ai-policy-congress-72004d-a1.

  133. HeldBy this siteGary Marcus

    Held record marcus-when-it-comes-to-ai-policy-congress-69cd67

    The anthropic review reads the claim differently.

  134. HeldBy this siteGary Marcus

    Held record marcus-when-it-comes-to-ai-policy-congress-678753

    mixed needs a supporting and a contradicting item of outcome evidence

  135. HeldBy this siteGary Marcus

    Held record marcus-when-it-comes-to-ai-policy-congress-5c747d

    The final-text check failed: citations_correct.

  136. Assessment publishedBy this siteGary Marcus

    If current trends continue, neither the US nor China will win the AI race.

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

    Assessment marcus-when-it-comes-to-ai-policy-congress-549063-a2.

  137. HeldBy this siteGary Marcus

    Held record marcus-when-it-comes-to-ai-policy-congress-4d6d84

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

  140. Assessment publishedBy this siteGary Marcus

    Congress should not preempt states, the only actors currently positioned to lead on AI regulation.

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

    Assessment marcus-when-it-comes-to-ai-policy-congress-2f185f-a1.

  141. Assessment publishedBy this siteGary Marcus

    Tech industry self-regulation has repeatedly failed the public.

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

    Assessment marcus-when-it-comes-to-ai-policy-congress-2a3cac-a1.

  142. HeldBy this siteGary Marcus

    Held record marcus-when-it-comes-to-ai-policy-congress-19d52b

    the conclusive event e1 is not outcome evidence in the packet

  144. Assessment publishedBy this siteGary Marcus

    US and Chinese AI results are converging because both use the same kinds of models, techniques, and data.

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

    Assessment marcus-when-it-comes-to-ai-policy-congress-013148-a1.

  145. Assessment publishedBy this siteGary Marcus

    Marcus doubts that language-model-driven robots like PaLM-SayCan should be built, even though they can sort of be made to work.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-bcda9d-a1.

  146. Assessment publishedBy this siteGary Marcus

    Google's PaLM-SayCan stitches superficial, incorrigible language models onto powerful and potentially dangerous humanoid robots.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-b5ab58-a1.

  147. Assessment publishedBy this siteGary Marcus

    The AI field currently has no compelling idea of how to solve the alignment problems of large language models.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-af3924-a1.

  149. Assessment publishedBy this siteGary Marcus

    DeepMind's list of 21 ethical and social problems with LLMs should be updated to cover the risks of putting LLMs in robots.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-8b233d-a1.

  150. Assessment publishedBy this siteGary Marcus

    In a general-purpose humanoid domestic robot, a 75% success rate is nowhere near good enough.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-895061-a1.

  151. Assessment publishedBy this siteGary Marcus

    Language model technology is likely to be even more problematic when used in robots.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-730ea2-2-a1.

  152. Assessment publishedBy this siteGary Marcus

    Large language models are powerful but reckless, like bulls in a china shop.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-6e0833-a1.

  153. Assessment publishedBy this siteGary Marcus

    As of August 2022, nobody can buy even a decently smart robot at any price.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-6d5a9b-a2.

  154. Assessment publishedBy this siteGary Marcus

    PaLM-SayCan's performance was around 75% at best, when the robot had a limited number of options.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-56bf7a-a1.

  155. HeldBy this siteGary Marcus

    Held record marcus-whats-wrong-with-googles-new-robot-55d126

    The anthropic review does not accept e1 as supports.

  156. Assessment publishedBy this siteGary Marcus

    The more actions a robot like PaLM-SayCan has to consider, the worse its performance is likely to be.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-3f36cb-a1.

  157. Assessment publishedBy this siteGary Marcus

    If large language models are put inside robots and misunderstand a request or miss its implications, they could cause major damage.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-2a0024-a1.

  158. HeldBy this siteGary Marcus

    Held record marcus-whats-wrong-with-googles-new-robot-202282

    hedged language is not scored, even on a plain reading

  159. Assessment publishedBy this siteGary Marcus

    PaLM-SayCan will not become a real-world product anytime soon.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-1ff17e-a1.

  160. Assessment publishedBy this siteGary Marcus

    LLM-driven robots will misunderstand people, and without firm checks some of those misunderstandings could lead to mayhem.

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

    Assessment marcus-whats-wrong-with-googles-new-robot-15223b-a1.

  163. Assessment publishedBy this siteGary Marcus

    AI policy may shape our long-term future more than tariff policy will.

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

    Assessment marcus-what-will-the-candidates-say-about-d224aa-a1.

  164. Assessment publishedBy this siteGary Marcus

    Market manipulation using AI-generated misinformation is likely to rise.

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

    Assessment marcus-what-will-the-candidates-say-about-8efdc3-a1.

  165. Assessment publishedBy this siteGary Marcus

    Citizens should now demand better from candidates on AI policy, or future generations may regret it.

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

    Assessment marcus-what-will-the-candidates-say-about-77c51f-a1.

  166. Assessment publishedBy this siteGary Marcus

    Presidential candidates should state their AI policies clearly and debate moderators should demand that clarity.

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

    Assessment marcus-what-will-the-candidates-say-about-6c6ac2-a1.

  168. Assessment publishedBy this siteGary Marcus

    A lot of shareholders could be angry, and unless there was a smoking gun, the litigation could be endless.

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

    Assessment marcus-what-went-down-at-openai-a-theory-2721f6-a1.

  169. Assessment publishedBy this siteGary Marcus

    According to The Information's reporting, it already looks like OpenAI's planned $86B valuation might not go through.

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

    Assessment marcus-what-went-down-at-openai-a-theory-1c8c3c-a1.

  170. Assessment publishedBy this siteGary Marcus

    In Marcus's view, the OpenAI board would not have fired Altman without some sort of documentation.

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

    Assessment marcus-what-went-down-at-openai-a-theory-0cd49f-a1.

  171. HeldBy this siteGary Marcus

    Held record marcus-what-washington-must-do-f9db3f

    The anthropic review does not accept e1 as supports.

  172. Assessment publishedBy this siteGary Marcus

    The US should govern AI deployments through an independent agency rather than opaque last-minute decisions by a small group.

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

    Assessment marcus-what-washington-must-do-f710f7-a1.

  173. HeldBy this siteGary Marcus

    Held record marcus-what-washington-must-do-ea664a

    The anthropic screen reads the claim differently.

  174. Assessment publishedBy this siteGary Marcus

    What Washington must do now is make the calls about which AI models might be risky in a clear and transparent way that doesn't reek of corruption.

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

    Assessment marcus-what-washington-must-do-d739a2-a1.

  175. Assessment publishedBy this siteGary Marcus

    The White House's Friday AI decision against Anthropic seems arbitrary and may even be corrupt.

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

    Assessment marcus-what-washington-must-do-d61d29-a1.

  176. Assessment publishedBy this siteGary Marcus

    The White House decision is a big win for China.

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

    Assessment marcus-what-washington-must-do-d5389b-a1.

  177. Assessment publishedBy this siteGary Marcus

    The idea of zero regulation of AI is now, rightly, dead.

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

    Assessment marcus-what-washington-must-do-8198e3-a1.

  178. Assessment publishedBy this siteGary Marcus

    The White House has acknowledged some AI models might be risky, so the debate is now only about details of regulation.

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

    Assessment marcus-what-washington-must-do-80b8fd-a1.

  179. Assessment publishedBy this siteGary Marcus

    There was no real justification for the speed at which the White House made a decision with such lasting consequences.

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

    Assessment marcus-what-washington-must-do-60633f-a1.

  180. Assessment publishedBy this siteGary Marcus

    The US government should act as the adult in the room and make AI decisions that are sound for the nation.

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

    Assessment marcus-what-washington-must-do-5f911a-a1.

  181. Assessment publishedBy this siteGary Marcus

    By its action the White House has vindicated Amodei's claim that some AI might need regulation and export controls.

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

    Assessment marcus-what-washington-must-do-47c03d-a1.

  182. Assessment publishedBy this siteGary Marcus

    The White House appears to have been looking for an excuse to harm Anthropic while ignoring larger consequences.

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

    Assessment marcus-what-washington-must-do-28a832-a1.

  183. Assessment publishedBy this siteGary Marcus

    Because the order targeted foreigners, it may cause a massive AI brain drain from the US.

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

    Assessment marcus-what-washington-must-do-090303-a1.

  184. HeldBy this siteGary Marcus

    Held record marcus-what-washington-must-do-0789ba

    The anthropic review does not accept e2 as supports.

  185. Assessment publishedBy this siteGary Marcus

    Governments must block unsafe AI deployments transparently, fairly, clearly, based on technical facts and without appearance of corruption.

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

    Assessment marcus-what-washington-must-do-04e3ef-a1.

  187. Assessment publishedBy this siteGary Marcus

    Laws should require machines to disclose that they are machines.

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

    Assessment marcus-what-was-60-minutes-thinking-in-that-f1275f-a1.

  188. Assessment publishedBy this siteGary Marcus

    A lot can be done to help with AI safety, but Marcus can see nothing that would absolutely guarantee it, and rushing ahead is creating risk.

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

    Assessment marcus-what-was-60-minutes-thinking-in-that-ed1996-a2.

  189. Assessment publishedBy this siteGary Marcus

    Seven years after Hinton's 2016 prediction, no radiologists have been replaced by machines.

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

    Assessment marcus-what-was-60-minutes-thinking-in-that-d9ffcc-a1.

  190. HeldBy this siteGary Marcus

    Held record marcus-what-was-60-minutes-thinking-in-that-c9e0ae

    e1 is dated 2023, which is not inside the period described (… to 2023-10-10); it can only be context; e2 is dated 2023, which is not inside the period described (… to 2023-10-10); it can only be context

  192. HeldBy this siteGary Marcus

    Held record marcus-what-was-60-minutes-thinking-in-that-b5c245

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

  193. Assessment publishedBy this siteGary Marcus

    A lot of people, Marcus included, still think Hinton is declaring victory for neural networks prematurely.

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

    Assessment marcus-what-was-60-minutes-thinking-in-that-a6a267-a1.

  194. HeldBy this siteGary Marcus

    Held record marcus-what-was-60-minutes-thinking-in-that-94d972

    e1 is dated 2023, which is not inside the period described (… to 2023-10-10); it can only be context; e2 is dated 2023, which is not inside the period described (… to 2023-10-10); it can only be context

  195. HeldBy this siteGary Marcus

    Held record marcus-what-was-60-minutes-thinking-in-that-8993da

    The anthropic review reads the claim differently.

  196. Assessment publishedBy this siteGary Marcus

    We are giving way too much authority to machines that we can't control.

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

    Assessment marcus-what-was-60-minutes-thinking-in-that-7e5e09-a1.

  197. Assessment publishedBy this siteGary Marcus

    We are not very close to machines that are more intelligent than humans.

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

    Assessment marcus-what-was-60-minutes-thinking-in-that-795b82-a1.

  199. Assessment publishedBy this siteGary Marcus

    Chatbots contain some information, but whatever is there is often both patchy and problematic.

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

    Assessment marcus-what-was-60-minutes-thinking-in-that-77bf3b-a1.

  200. HeldBy this siteGary Marcus

    Held record marcus-what-was-60-minutes-thinking-in-that-6442a3

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