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

Claims checked against evidence, page 22 of 31

Claims by Gary Marcus and Ed Zitron. Deadlines are judged as of 2026-10-29.

Claims checked against evidence (3100)

Each of these claims was checked against dated evidence.

Page 22 of 31: claims 2101 to 2200 of 3100, ordered by the date first said, newest first. To find a claim by its words, use search.

  1. Assessment: ContradictedSaid

    Ed Zitron

    “Along with $3.5 billion from Saudi Arabia’s Public Investment fund and over $8 billion from Softbank, the latter a secondary market sale, it also raised (equity and debt financing respectively) from Goldman Sachs and Morgan Stanley — two parties I do not believe are going to be willing to subsidize generative AI.”

    What happened: Both named banks participated in OpenAI's executed credit facility, demonstrating willingness to finance the business. The report's commercial interest terms do not establish a concessionary subsidy. (CNBC, )

  2. Assessment: UnresolvedSaid

    Ed Zitron

    “It would require a way of expanding what OpenAI can sell to address the entirety of corporate America, which would require use cases that I do not believe generative AI is capable of meeting, like automating chunks of the economy rather than bankrupting freelance designers and copy-editors.”

  3. Assessment: UnresolvedSaid

    Ed Zitron

    “As a result, expectations are higher of what future generations of GPT might do, making it inevitable that the company will disappoint investors and customers.”

  4. Assessment: UnresolvedSaid

    Ed Zitron

    “here’s the crux of the matter: generative AI is a product with no mass-market utility - at least on the scale of truly revolutionary movements like the original cloud computing and smartphone booms - and it’s one that costs an eye-watering amount to build and run.”

  5. Assessment: SupportedSaid

    Ed Zitron

    “Pre-AGI in this case refers to anything OpenAI has ever developed, as it has yet to develop AGI and has yet to get past the initial "chatbot" stage of its own 5-level system of evaluating artificial intelligence.”

    What happened: The report places OpenAI just below level 2. Approaching that level does not establish that it had passed the initial chatbot stage. (The Decoder, )

  6. Assessment: UnresolvedSaid

    Ed Zitron

    “I am simply saying that for OpenAI to grow to the $10+ billion revenue a year it needs to survive, it would need to replace entire chunks of the labor force.”

  7. Assessment: UnresolvedSaid

    Ed Zitron

    “Raise either multiple rounds, or the largest funding round ever raised by any company, and then have to keep doing so in perpetuity.”

  8. Assessment: SupportedSaid

    Ed Zitron

    “And OpenAI must grow, because $3.5 billion to $4.5 billion a year in revenue is simply not enough to keep this company going.”

    What happened: The article reports estimated 2024 operating costs of $8.5 billion against estimated revenue of $3.5–4.5 billion. Both ends of the revenue range fall below the estimated costs, supporting the operating-coverage reading. (The Decoder, )

  9. Assessment: UnresolvedSaid

    Ed Zitron

    “I see no signs that the transformer-based architecture can do significantly more than it currently does.”

  10. Assessment: UnresolvedSaid

    Ed Zitron

    “If it succeeds — which I add would require potentially entirely new branches of psychology and mathematics to do so, as we humans barely understand our brains — that would be a huge technological achievement that still wouldn't turn things around.”

  11. Assessment: UnresolvedSaid

    Ed Zitron

    “though I believe OpenAI's continued existence is necessary to keep companies interested/invested in the industry at all.”

  12. Assessment: UnresolvedSaid

    Ed Zitron

    “Conversely, OpenAI has devoured no monopoly, and the product category it created — the only one it is really part of — is one entirely subsidized by venture capital and Microsoft.”

  13. Assessment: SupportedSaid

    Gary Marcus

    “Consequently, any “reasoning” or “planning” in LLMs is hit or miss, viable if the particulars of a given situation are close enough to what is in the training set, but highly vulnerable if they are not.”

    What happened: Reasoning accuracy falls substantially and consistently on unfamiliar variants of the same tasks. Performance is still nontrivial, which fits 'hit or miss' rather than total failure. (Wu et al., MIT/Boston University (arXiv; NAACL 2024), )

  14. Assessment: UnresolvedSaid

    Gary Marcus

    “People are starting to realize all of the above.”

  15. Assessment: SupportedSaid

    Gary Marcus

    “Much as I admire the new pair of systems, they do have a serious shortcoming and one that is not unfamiliar. The systems both rely on a kind of cheat on the input: trained human coders translate all the Olympiad's input sentences into mathematical form.”

    What happened: The primary paper explicitly confirms expert manual formalization of AlphaProof's competition inputs. Its publication is later, but the reported event is the July 2024 evaluation. (Nature, )

  16. Assessment: UnresolvedSaid

    Gary Marcus

    “That said, I am certain that the impending collapse won’t lead to the absolute death of AI. There is too much at stake.”

  17. Assessment: ContradictedSaid

    Gary Marcus

    “I fully expect that the generative AI bubble will begin to burst within the next 12 months, for many reasons:”

    What happened: EY reports $49.2 billion of first-half generative AI VC investment versus $44.2 billion for all of 2024. It also reports nearly 25% fewer transactions than in the preceding half-year, a material sign of concentration that qualifies the expansion. (EY Ireland, )

  18. Assessment: UnresolvedSaid

    Gary Marcus

    “There can be no path to AGI without neurosymbolic AI.”

  19. Assessment: UnresolvedSaid

    Gary Marcus

    “Nobody has a moat”

  20. Assessment: SupportedSaid

    Gary Marcus

    “Nonetheless, GDM's new work is powerful proof of concept (pun intended!). Neural networks and symbol systems can be profitably combined, contra Hinton and LeCun.”

    What happened: The developer's account connects the actual IMO proofs to a transformer that proposes Lean tactics, search over proof states and automatic formal verification. It explicitly identifies the described work as ending in July 2024. (Julian Schrittwieser, AlphaProof coauthor, )

  21. Assessment: UnresolvedSaid

    Gary Marcus

    “The neural network approach known as generative AI — especially its poster chatbot ChatGPT – has captured the world’s imagination, but will soon flame out.”

  22. Assessment: SupportedSaid

    Gary Marcus

    “Hallucinations and boneheaded errors of reasoning and what I like to call discomprehensions are, as far as I can tell, an inevitable byproduct of LLMs; certainly they have yet to disappear, despite many promises to the contrary.”

    What happened: In mid-2024, LLM products whose providers claimed they 'avoid' hallucinations still hallucinated on 17% to 34% or more of test queries. (Stanford HAI / RegLab, )

  23. Assessment: ContradictedSaid

    Gary Marcus

    “The current approach has reached a plateau”

    What happened: User comparisons collected June 10–25 show newer frontier models outperforming preceding models on image-containing conversations. Preference measurements are imperfect, but this is independent evidence of improvement within the target quarter. (LMSYS Org, )

  24. Assessment: UnresolvedSaid

    Gary Marcus

    “So for the most part, we’ve been stuck in a world in which all the major tech companies are building exactly the same thing (bigger and bigger LLMs based on more data), achieving almost exactly the same results (a passel of GPT-4 level models, hardly different from each other, all struggling with hallucinations and boneheaded errors), and hardly any investment in anything else.”

  25. Assessment: UnresolvedSaid

    Gary Marcus

    “There is no killer app”

  26. Assessment: UnresolvedSaid

    Gary Marcus

    “The biggest intrinsic failings of generative AI have to do with reliability, in a way that I believe can never be solved, given their inherent nature.”

  27. Assessment: UnresolvedSaid

    Gary Marcus

    “Long-term adoption (vs mere experimentation) has been modest, there is no significantly better tech now than there was in January; GPT-5 is still a no show.”

  28. Assessment: SupportedSaid

    Gary Marcus

    “Long-term adoption (vs mere experimentation) has been modest, there is no significantly better tech now than there was in January; GPT-5 is still a no show.”

    What happened: The official July release history records GPT-4o mini, not GPT-5. This supports the chronology alongside explicit reporting that GPT-5 remained forthcoming; this entry alone would not prove nonrelease. (OpenAI, )

  29. Assessment: UnresolvedSaid

    Gary Marcus

    “Customers can easily switch to other providers, almost anytime they want, and for month there had been more and more alternatives, including Anthropic and Google, at more and more attractive prices, forcing OpenAI itself to cut prices just a few days ago.”

  30. Assessment: UnresolvedSaid

    Gary Marcus

    “LLMs have just became exactly the commodity I predicted they would become, at the lowest possible price.”

  31. Assessment: UnresolvedSaid

    Gary Marcus

    “Obviously, their only hope is to raise more cash, and they will certainly try.”

  32. Assessment: ContradictedSaid

    Ed Zitron

    “Even though it doesn't really automate anything.”

    What happened: Intercom reported that deployed Fin systems resolved customer questions directly. The existence of successful bounded tasks contradicts 'anything', although the reported percentage requires qualification. (Intercom, )

  33. Assessment: ContradictedSaid

    Ed Zitron

    “Here's the thing, I get that that might be what they're pushing toward, but generative AI isn't even automation at this point.”

    What happened: Intercom reported actual customer deployments resolving queries directly through Fin. This supports the existence of bounded task automation, although the reported percentage is not an independently verified success rate. (Intercom, )

  34. Assessment: UnresolvedSaid

    Ed Zitron

    “Because right now the thing I've been on about with generative AI is on top of not being super useful, it's so unprofitable.”

  35. Assessment: UnresolvedSaid

    Gary Marcus

    “And leaving more and more code writing to generative AI, which grasps syntax but not meaning, as a new benchmark from Stephen Wolfram shows, is not the answer, either.”

  36. Assessment: SupportedSaid

    Gary Marcus

    “As Ernie Davis and I pointed out in Rebooting AI, five years ago, part of the reason we are struggling with AI in complex AI systems is that we still lack adequate techniques for engineering complex systems. That hasn’t really changed.”

    What happened: The researchers explain that model scale and structure make established software testing and verification methods difficult to apply. They connect that methodological limitation to the difficulty of ensuring deployment robustness. (DeepMind, )

  37. Assessment: UnresolvedSaid

    Ed Zitron

    “As these organizations push their engineers harder, said engineers will turn to AI-generated code, poisoning codebases with insecure and buggy code as companies shed staff to keep up with Wall Street’s demands in ways that I’m not sure people are capable of understanding.”

  38. Assessment: UnresolvedSaid

    Ed Zitron

    “they're allowed to get away from the very obvious problem: that Large Language Models are peaking, will not solve the kind of complex problems that actually matter, and OpenAI (and other LLM companies) are being allowed to accumulate money and power in a way that's allowed them to do actual damage in broad daylight.”

  39. Assessment: UnresolvedSaid

    Ed Zitron

    “The media seems nigh-on incapable of accepting that generative AI is a big, stupid, costly and environmentally-destructive bubble, to the point that they'll happily accept marketing slop and vague platitudes about how big something that's already here will be in the future based on a remarkable lack of proof.”

  40. Assessment: UnresolvedSaid

    Ed Zitron

    “The problem with stories like this is that they suggest that OpenAI is working on Artificial General Intelligence, or something that left unchecked could somehow destroy society, as opposed to what it’s actually working on — increasingly faster iterations of a Large Language Model that's absolutely not going to do that.”

  41. Assessment: UnresolvedSaid

    Ed Zitron

    “The problem is that these pieces aren’t about breakthroughs — they’re marketing collateral dressed up as technological progress, superficially-exciting yet hollow on the inside, which almost feels a little too on the nose.”

  42. Assessment: UnresolvedSaid

    Ed Zitron

    “I guarantee you Anthropic, Google and Meta are all working on something doing exactly the same thing, because what they are describing is something that limits every single Large Language Model due to the very nature of how they’re built, something Reuters even acknowledges by adding that “this is something that has eluded AI models to date.””

  43. Assessment: ContradictedSaid

    Ed Zitron

    “To reach level 3 (whatever “Agents, systems that can take actions” means) you’d need a level of sentience dependent on entirely new forms of mathematics.”

    What happened: The account describes Claude Sonnet 3.7 using search, notes, Slack, and price-setting tools over a month, with humans providing physical services. I infer from this documented implementation that sustained action-taking did not require the asserted new mathematical foundations. (Anthropic, )

  44. Assessment: UnresolvedSaid

    Ed Zitron

    “While there's tons of evidence that they're impossible to fix, surely a safety-minded culture would be one that sought to fix or mitigate the most obvious and dangerous problem of Large Language Models.”

  45. Assessment: UnresolvedSaid

    Ed Zitron

    “In essence, on top of generative AI not having any killer apps, not meaningfully increasing productivity or GDP, not generating any revenue, not creating new jobs or massively changing existing industries, it also requires America to totally rebuild its power grid, which Janous regrettably adds the US has kind of forgotten how to do.”

  46. Assessment: UnresolvedSaid

    Ed Zitron

    “I see no situation where OpenAI and Anthropic continue to iterate on Large Language Models in perpetuity, as at some point Microsoft, Amazon and Google decide (or are forced to decide) that cloud compute welfare isn't a business model.”

  47. Assessment: ContradictedSaid

    Ed Zitron

    “And yes, that sound you hear is the slow deflation of the bubble I've been warning you about since March.”

    What happened: Their account of PitchBook data reports record generative-AI investment in Q2. The same article describes weaker US seed activity and cooler enthusiasm, so the contradiction concerns broad financial contraction. (Aspiring for Intelligence — Sabrina Albert and Vivek Ramaswami, )

  48. Assessment: UnresolvedSaid

    Ed Zitron

    “Basically, there's going to be a moment that spooks a venture capital firm into pushing one of its startups to sell, or the sudden, unexpected yet very obvious collapse of a major player.”

  49. Assessment: UnresolvedSaid

    Ed Zitron

    “All of this excitement, every second of breathless hype has been built on this idea that the artificial intelligence industry – led by generative AI – will somehow revolutionize everything from robotics to the supply chain, despite the fact that generative AI is not actually going to solve these problems because it isn't built to do so.”

  50. Assessment: UnresolvedSaid

    Ed Zitron

    “The reason I'm suddenly bringing up superintelligences — or AGI (artificial general intelligence) — is because throughout every defense of generative AI is a deliberate attempt to get around the problem that generative AI doesn't really automate many tasks.”

  51. Assessment: UnresolvedSaid

    Ed Zitron

    “Generative AI, as I said back in March, is peaking, if it hasn't already peaked.”

  52. Assessment: ContradictedSaid

    Ed Zitron

    “It cannot do much more than it is currently doing, other than doing more of it faster with some new inputs.”

    What happened: The public evaluation reported 97.8% success on basic Blocksworld and 52.8% on its obfuscated version, where earlier tested LLMs remained below 5%. These are substantive planning gains; the event date denotes publication of the demonstration, since exact test-run dates are not supplied. (Valmeekam, Stechly, Gundawar and Kambhampati / arXiv, )

  53. Assessment: UnresolvedSaid

    Ed Zitron

    “The jobs market isn't going to change because of generative AI, because generative AI can't actually do many jobs, and it's mediocre at the few things that it's capable of doing.”

  54. Assessment: UnresolvedSaid

    Ed Zitron

    “Or, put another way, each additional step becomes increasingly (and exponentially) more expensive to take.”

  55. Assessment: SupportedSaid

    Ed Zitron

    “For Goldman to suddenly turn on the AI movement suggests that it’s extremely anxious about the future of generative AI, with almost everybody agreeing on one core point: that the longer this tech takes to make people money, the more money it's going to need to make.”

    What happened: The investor's model illustrates the recurring recovery burden from continued AI infrastructure spending. Its numerical estimate is not treated as audited industry profit or as a figure claimed by Zitron. (Sequoia Capital — David Cahn, )

  56. Assessment: UnresolvedSaid

    Ed Zitron

    “It's been eighteen months since this bubble inflated, and since then very little has actually happened involving technology doing new stuff, just an iterative exploration of the very clear limits of what an AI model that generates answers can produce, with the answer being "something that is, at times, sort of good."”

  57. Assessment: UnresolvedSaid

    Ed Zitron

    “Yet, I have a counterpoint: no it doesn't.”

  58. Assessment: ContradictedSaid

    Ed Zitron

    “It isn’t getting much more efficient.”

    What happened: The paper reports stronger model performance with 93.3% less attention-cache memory than DeepSeek 67B. Its deployment tests report 5.76 times the generation throughput on an eight-H800 node using the earlier service's workload distribution, demonstrating substantial resource-efficiency gains. (DeepSeek-AI / arXiv, )

  59. Assessment: UnresolvedSaid

    Ed Zitron

    “While it's a useful efficiency tool, said efficiency is based off of a technology that is extremely expensive, and I believe that at some point AI companies like Anthropic and OpenAI will have to increase prices — or begin to collapse under the weight of a technology that has no path to profitability.”

  60. Assessment: MixedSaid

    Ed Zitron

    “Behind the scenes, generative AI isn’t moving the needle”

    What held: Named executives described inaccurate answers and substantial preparation work, with Juniper not ready for production deployment. This supports the narrower contention that some enterprises were struggling to obtain useful results. (The Wall Street Journal, reproduced by AI Incident Database, )

    What did not: Klarna reported a material reduction in repeat inquiries from an operating generative AI assistant. This is a realized operational benefit, although company-reported and not a measure of net financial return. (Klarna, )

  61. Assessment: UnresolvedSaid

    Ed Zitron

    “even if their models are all extremely similar, as they’re all trained on the same quickly-dwindling amount of training data.”

  62. Assessment: UnresolvedSaid

    Ed Zitron

    “They don’t care that the output is mediocre, that the product is unprofitable, that there is a quickly-approaching wall that generative AI can’t leap over as it runs out of training data, or that the transformer-based architecture of Large Language Models has hard limitations that are impossible to overcome.”

  63. Assessment: UnresolvedSaid

    Ed Zitron

    “The deep poison of the Shareholder Supremacy — the nihilistic intention of contorting companies to please analysts and investors — is one that will uniquely punish the tech industry, just like it did to General Electric.”

  64. Assessment: UnresolvedSaid

    Ed Zitron

    “Google, Microsoft and Meta have bet their futures on generative AI, cramming it into their products with little regard for utility, all to show the markets that they’re futuristic rocket ship growth companies”

  65. Assessment: ContradictedSaid

    Ed Zitron

    “Generative AI is hitting a wall, and any future improvements will be marginal, and cost billions to achieve.”

    What happened: ARC Prize reports 75.7% and 87.5% for the o3 preview configurations and contrasts these with GPT-4o's 5% performance in 2024. The evaluator describes a substantial advance, while disclosing benchmark-specific training and increased inference computation. (ARC Prize, )

  66. Assessment: UnresolvedSaid

    Ed Zitron

    “Microsoft is run by Satya Nadella, another MBA that has, in his tenure, overseen layoffs of over 30,000 people, and pushed his company deep into the deeply unprofitable and unsustainable world of generative AI.”

  67. Assessment: UnresolvedSaid

    Gary Marcus

    “Neurosymbolic AI has long been an underdog; in the end I expect it to come from behind and be essential.”

  68. Assessment: UnresolvedSaid

    Gary Marcus

    “The markers that our next President puts down—about how much liability and responsibility companies will have, and about how the government keeps up with a rapidly changing technology— will have lasting impact.”

  69. Assessment: UnresolvedSaid

    Gary Marcus

    “Returns are starting to diminish.”

  70. Assessment: UnresolvedSaid

    Gary Marcus

    “With the truth that LLMs can’t be trusted now becoming more widely known, the big AI leaders have switched their focus to long term promises that are impossible to verify and even more grandiose than “we can help your business in unprecedented ways” (which has turned out to be mostly untrue).”

  71. Assessment: SupportedSaid

    Gary Marcus

    “As of today, I am more confident than ever that GPT-5 won’t land this year.”

    What happened: OpenAI introduced GPT-5 on August 7, 2025. A model first introduced in August 2025 was not released in 2024. (OpenAI, )

  72. Assessment: SupportedSaid

    Gary Marcus

    “As of today, I am more confident than ever that GPT-5 won’t land this year.”

    What happened: OpenAI's initial GPT-5 announcement is dated August 7, 2025, and describes rollout beginning that day. This places its public arrival after the forecast's 2024 deadline. (OpenAI, )

  73. Assessment: UnresolvedSaid

    Gary Marcus

    “In case you haven’t noticed, GPT-5 has come to seem to be a little like that. People keep on trying project its arrival, and its arrival keeps on slippin’ into the future.”

  74. Assessment: UnresolvedSaid

    Gary Marcus

    “Nobody to my knowledge has kept systematic track of the predictions, but I took a quick and somewhat random look at X and had no trouble finding many predictions, going back to 2023, almost always optimistic.”

  75. Assessment: ContradictedSaid

    Ed Zitron

    “While Google has desperately tried (and failed) to convince Wall Street that it’s building and selling the future”

    What happened: Goldman Sachs's reported assessment after I/O is direct evidence that a prominent Wall Street institution accepted Google's AI leadership and product-integration pitch. The original analyst note was not accessible, so this assessment relies on reporting. (Yahoo Finance, )

  76. Assessment: UnresolvedSaid

    Ed Zitron

    “Yet Apple’s deal doesn’t even appear to pay OpenAI for using ChatGPT, meaning that OpenAI’s deal will immediately lose them money from the second iOS 18 launches.”

  77. Assessment: UnresolvedSaid

    Gary Marcus

    “Others are bailing now, too. Or at least trying to subtly alter their positions, committing less to the unrealistic.”

  78. Assessment: SupportedSaid

    Gary Marcus

    “Exhibit A: OpenAI’s CTO Mira Murati just publicly acknowledged what I long suspected: there is no mind blowing GPT-5 behind the scenes as of yet.”

    What happened: The contemporaneous report attributes a modest internal capability lead to OpenAI's CTO. This supports the ordinary qualitative reading, although it does not independently measure private models or identify one as GPT-5. (The Decoder, )

  79. Assessment: UnresolvedSaid

    Gary Marcus

    “LLMs are not AGI, and (on their own) never will be; scaling alone was never going to be enough.”

  80. Assessment: UnresolvedSaid

    Ed Zitron

    “Generative artificial intelligence is not going to become the incredible supercomputer that Altman has regularly promised.”

  81. Assessment: UnresolvedSaid

    Ed Zitron

    “When the deck of cards collapses, he’ll find something to blame it on — market forces, a lack of energy breakthroughs, or “unfortunate economic forces” — and remain a billionaire capable of doing anything he wants.”

  82. Assessment: SupportedSaid

    Ed Zitron

    “As things get desperate, Altman will use the only approach he knows: sheer force of will. He will push OpenAI to grow and sell into as many industries as possible, a specious hype-man selling to other specious hype-men desperate to connect their company to an AI bubble inflated by Altman himself.”

    What happened: The article reports that Altman declared a code red to avoid falling behind competitors and redirected work. This meets the stated reading of serious business pressure, though it does not establish insolvency. (MacRumors, )

  83. Assessment: UnresolvedSaid

    Ed Zitron

    “He will push OpenAI to grow and sell into as many industries as possible, a specious hype-man selling to other specious hype-men desperate to connect their company to an AI bubble inflated by Altman himself.”

  84. Assessment: SupportedSaid

    Gary Marcus

    “In many ways, it is much weaker than what Altman and I talked about at the Senate last year.”

    What happened: Marcus proposed FDA-like review before deployment. The same transcript records Altman's proposed licensing agency, license revocation, safety standards, and independent audits; it establishes the comparison rather than proving Marcus's later assessment by repetition. (U.S. Senate / Government Publishing Office, )

  85. Assessment: UnresolvedSaid

    Gary Marcus

    “Anyone who can afford that can easily fill out some paperwork and run a few tests, just like I can pay a few bucks to get a driver’s license. It’s the least they can do.”

  86. Assessment: ContradictedSaid

    Gary Marcus

    “He doesn’t specify which actual restrictions in the proposal bill would “stifle AI models”.”

    What happened: Ng explicitly names safety assessments and shutdown capabilities. In the next paragraph, he focuses on the hazardous-capability designation and potential downstream liability, then connects that burden to inhibited model development, especially open source. (DeepLearning.AI, )

  87. Assessment: SupportedSaid

    Gary Marcus

    “Two weeks later, it has already come out that the database that indexes the whole thing is in plain text (except on systems that are explicitly set to encrypt everything). That’s bananas. Keystone Kops level security.”

    What happened: The preserved June documentation describes the unencrypted database and demonstrates extracting and searching its screenshot-derived text. Its June 6 example falls within the target period; June 8 is the documentation's stated FAQ update date. (Alexander Hagenah, )

  88. Assessment: UnresolvedSaid

    Gary Marcus

    “Every company that would be subject to the stronger set of regulations is valued in the billions.”

  89. Assessment: UnresolvedSaid

    Gary Marcus

    “It can do some things better, some worse, very few reliably.”

  90. Assessment: UnresolvedSaid

    Gary Marcus

    “The double Y-axis makes no sense, and presupposes its own conclusion.”

  91. Assessment: MixedSaid

    Gary Marcus

    “The graph assumes that all continued progress will be same — ignoring qualitatively unsolved problems (hallucinations, planning, reasoning etc), the fact that synthetic data may not be as useful as natural data, limits on then-possible energy distribution and compute, etc.”

    What held: Supports the trend-continuation component, not the alleged omission of obstacles. (Leopold Aschenbrenner, Situational Awareness, )

    What did not: The surrounding discussion explicitly allows data constraints to halt progress without new methods. (Leopold Aschenbrenner, Situational Awareness, )

  92. Assessment: SupportedSaid

    Gary Marcus

    “For one thing, the bad press about Sam Altman and OpenAI, who once seemingly could do no wrong, just keeps coming.”

    What happened: This critical newsletter discusses the voice dispute, safety-team departures, and restrictive employee agreements. The event being assessed is publication of critical coverage; available_at conservatively uses the displayed modification date. (The Guardian, )

  93. Assessment: UnresolvedSaid

    Gary Marcus

    “In 2019 I pointed out through a series of examples on Twitter with GPT-2 that LLMs were subject to essentially the same critique: greedy with respect to data, uninterpretable, unreliable as reasoners, subject to hallucinations, etc, that they lacked world models (“they don't develop robust representations of *how events unfold over time*), and that they would be inadequate on their own.”

  94. Assessment: UnresolvedSaid

    Gary Marcus

    “Nowadays we both (unlike many other people) stress the absolutely essential nature of common sense, physical reasoning and world models, and the failure of current architectures to handle those well.”

  95. Assessment: UnresolvedSaid

    Ed Zitron

    “They both went through yCombinator, and both of them have sold products that don't do what they said it would based on what it might do in the future.”

  96. Assessment: UnresolvedSaid

    Ed Zitron

    “The Rot-Com bubble will inevitably pop, but until it does, billions will be plowed into building hardware and software to inflate the Rot-Com bubble in the hopes that the growth-at-all-costs party will never end.”

  97. Assessment: SupportedSaid

    Ed Zitron

    “Despite being an inherently AI-powered feature, it can’t distinguish between ordinary computer usage and genuinely sensitive information — like passwords, health information, or trade secrets that must be protected at all costs.”

    What happened: The contemporary reproduction of Microsoft's own FAQ explicitly says that the preview did not hide sensitive information. The surrounding FAQ lists manual and application-based exclusions, which do not amount to automatic sensitivity recognition. (Gibson Research Corporation, Security Now!, )

  98. Assessment: UnresolvedSaid

    Ed Zitron

    “Generative AI is not going to change the world, and while artificial intelligence itself might, it isn't going to do so in the hands of Sam Altman and OpenAI, or Anthropic.”

  99. Assessment: UnresolvedSaid

    Ed Zitron

    “The answer is simple: it isn't coming.”

  100. Assessment: UnresolvedSaid

    Gary Marcus

    “Worse, we are no closer to a solution to the alignment problem now than we were then.”