Changes, page 19 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 19 of 37: changes 3601 to 3800 of 7383, newest first. Dates are when the change was recorded.

2026-10-29
Assessment publishedBy this siteGary Marcus
Taking the Salesforce report together with the Apple paper, it is clear that the current tech is not to be trusted.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-replies-to-the-viral-apple-e04f47-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Over time, scores of papers will reinforce the Apple paper's findings on reasoning model limitations.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-replies-to-the-viral-apple-dbeb5a-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Reaching AGI will require doing better than current LLMs.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-replies-to-the-viral-apple-db6501-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-seven-replies-to-the-viral-apple-b58cc0
The anthropic screen finds the claim checkable, so it must be graded.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs retrieving code without conceptual understanding help less on new problems and dynamically changing environments.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-replies-to-the-viral-apple-a5c0a8-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
If people like Sam Altman are sweating, it is because they should be.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-replies-to-the-viral-apple-9721ba-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-seven-replies-to-the-viral-apple-8c90aa
The openai screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
People are finally starting to pay attention to poor generalization, one of the two biggest Achilles' heels of generative AI, and to appreciate its significance.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-replies-to-the-viral-apple-829ca9-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
That the systems can solve the puzzles with code, which is true in some cases, is a huge win for neurosymbolic AI, given that they can't reliably solve the puzzles without code and code is symbolic.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-replies-to-the-viral-apple-76d19a-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
AI should integrate neural networks with symbolic algorithms and representations.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-replies-to-the-viral-apple-59d7cf-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Many more examples of LLM/LRM failures at applying algorithms, like those in the Apple paper, will be uncovered.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-replies-to-the-viral-apple-367767-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Because some model might be big enough for a task of one size yet fail on the next size or on a slightly different task, one is left having to test everything all the time with little guarantee of anything, making it all a crapshoot.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-replies-to-the-viral-apple-31b05e-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
OpenAI absorbs essentially none of the negative externalities of its products, such as misinformation, cybercrime, and fake books.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-fc0fed-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Sam Altman's request for $7 trillion shows he has jumped the shark.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-f448c7-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Scaling AI insanely fast risks failure that could damage the whole field of AI.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-f1bec4-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
A project that costs $7 trillion will inevitably be seen as too big to fail, and any bailout could easily shatter the world's economy.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-df3cf8-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Large language models are now making scams like fake books easy, and companies like OpenAI seem content to let society bear all the costs.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-beeb12-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
So far GenAI has been mostly promise rather than delivery, and there is no argument yet that it is actually worth tearing up the globe.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-a313d0-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
AI research should not be abandoned altogether, but for now the focus should be primarily on safety and innovative new approaches rather than rushing to scale up existing, problematic architectures.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-9830c8-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
We should all say no to reckless AI expansionism unless and until tech leaders can show persuasively that the technology is mature enough that such expansion would demonstrably yield enormous net benefit.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-800844-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
GenAI's future energy needs will probably exceed the power consumption of Germany, especially if models keep growing.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-798bd1-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The collapse of a $7 trillion AI project could trigger a worldwide financial depression worse than the 2007-2008 crisis.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-6a4eee-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-seven-reasons-why-the-world-should-352f92
The anthropic review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
The AI alignment problem is not remotely solved.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-2251b1-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-seven-reasons-why-the-world-should-215ca9
The openai screen doubts the attribution.
2026-10-29
Assessment publishedBy this siteGary Marcus
If competition over resources for AI chips became acute enough, actual war could follow.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-139f51-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Taking a real chance at wrecking both society and the environment on little more than vague speculation makes no sense.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-why-the-world-should-07d693-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-seven-reasons-i-wouldnt-count-google-ea7188
The anthropic screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
Hassabis and his replacement Kavukcuoglu will keep working together at Google.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-i-wouldnt-count-google-e611e9-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Google has strong distribution through Android, Mail, Search, Docs and YouTube.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-i-wouldnt-count-google-d94efe-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Google may outlast OpenAI and Anthropic.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-i-wouldnt-count-google-bf4c78-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
If things turn out as he expects, with the LLM market a multi-way tie with relatively modest profits, Anthropic and OpenAI don't have that much else to fall back on and may never earn out their investments.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-i-wouldnt-count-google-bc6e39-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-seven-reasons-i-wouldnt-count-google-8bc68d
The openai screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
Google definitely doesn't look unassailable as it once did, and may never be as strong as it once was.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-i-wouldnt-count-google-751e65-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
It is premature to count Google out.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-i-wouldnt-count-google-751e65-2-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Kavukcuoglu's low name recognition may hurt Google short-term, but his sobriety may prove crucial long-term.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-i-wouldnt-count-google-5bbb07-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-seven-reasons-i-wouldnt-count-google-37a933
The anthropic screen finds the claim checkable, so it must be graded.
2026-10-29
Assessment publishedBy this siteGary Marcus
Google's competitors have problems: OpenAI's reputation is declining, Anthropic is in conflict with the administration, and Microsoft and Amazon lack comparable talent.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-i-wouldnt-count-google-35ea9b-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Google made $402 billion revenue and $132 billion profit last year and has vastly more cash than OpenAI and Anthropic.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-reasons-i-wouldnt-count-google-35693d-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs produce plausible falsehoods because they mimic existing sentences like autocomplete rather than checking what they say against the world.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-lies-in-four-sentences-c392e1-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs have not solved the hallucination problem.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-lies-in-four-sentences-99c9db-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
A recent Microsoft Bing answer about the House Speaker contained at least seven falsehoods.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-lies-in-four-sentences-6dc2e4-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The field should commit to building AI that can distinguish facts from fiction.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-lies-in-four-sentences-5eb1e0-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-seven-lies-in-four-sentences-55debe
The final-text check failed: rationale_supported_by_evidence, citations_correct.
2026-10-29
Assessment publishedBy this siteGary Marcus
Bullshit was and still is the fundamental product of LLMs.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-lies-in-four-sentences-18fbff-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The worst part of Bing's false answer is not that every sentence contains at least one falsehood but that the whole thing sounds detailed and real, so that to someone who wasn't following matters closely it might well all sound plausible.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-seven-lies-in-four-sentences-0da33b-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
People should never trust today's chatbots.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-serious-medical-error-from-perplexitys-2e48be-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sentient-ai-for-the-love-of-darwin-f933bc
The anthropic screen doubts the attribution.
2026-10-29
Assessment publishedBy this siteGary Marcus
AI is subject to the risk of unintended consequences, possibly more than other technologies, and expanding its capabilities increases that risk.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sentient-ai-for-the-love-of-darwin-c49b1c-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
We should not blithely assume that anything we might invent in AI can be dealt with.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sentient-ai-for-the-love-of-darwin-9003ea-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Questions about consciousness probably cannot yet really be answered, given current science and technology, so we must settle for understanding the arguments clearly.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sentience-and-ai-a-dialog-between-ee2a57-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The argument over whether AI systems are sentient may well go on for a long time, evolving as the technology advances.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sentience-and-ai-a-dialog-between-b1a32c-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Even in the worst case, AI probably will not lead to every human on the planet dying.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-self-modern-post-parody-e75985-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Big tech AI efforts are regularly and predictably failing in the same way, with ever larger amounts of money spent.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-self-modern-post-parody-a67436-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Consumers are starting to catch on, apparently to the AI failures and hype the post has just described, though the post does not say exactly what they are catching on to.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-self-modern-post-parody-8b47a3-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-self-modern-post-parody-64906a
The openai screen doubts the attribution.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-self-modern-post-parody-309dea
e1 measures something else (metric net loss is not net profit (net income)); it can only be context unless it is derived with stated inputs; e2 measures something else (metric net loss is not net profit (net income)); it can only be context unless it is derived with stated inputs; e3 measures something else (metric net loss is not net profit (net income)); it can only be context unless it is derived with stated inputs
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-self-modern-post-parody-29759e
The anthropic review does not accept e3 as contradicts.
2026-10-29
Assessment publishedBy this siteGary Marcus
At the end of 2027 it will still be easy to find clear counterexamples where LLM-based generative AI fails at compositionality, because of limitations inherent in LLMs.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scott-alexanders-misleading-victory-e7b983-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scott-alexanders-misleading-victory-a5c2d6
The openai screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
It is not reasonable to conclude from occasional success on one sort of publicly discussed set of prompts that GPT has mastered compositionality in general.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scott-alexanders-misleading-victory-99f09c-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scott-alexanders-misleading-victory-6b05c6
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
It is an enormous, unjustified leap from 'the systems can now solve some compositionality problems' to 'the systems have mastered compositionality'.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scott-alexanders-misleading-victory-5c5f30-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scott-alexanders-misleading-victory-4df734
The openai review reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
Publicly circulated prompts may become part of training data (including via data augmentation by subcontractors), making success on them unreliable evidence of general ability.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scott-alexanders-misleading-victory-4c0f5a-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Once AI truly reaches AGI, finding compositionality counterexamples will be much harder than it is now.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scott-alexanders-misleading-victory-3c4d1e-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
In less than an hour, Marcus found half a dozen clear counterexamples in which ChatGPT 4o failed with respect to compositionality, failures evident even to a teenager.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scott-alexanders-misleading-victory-303f61-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scoop-what-former-employees-of-openai-f93cf1
The anthropic screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
California's SB-1047, already only a small step toward external AI governance, has been watered down.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scoop-what-former-employees-of-openai-d30c92-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
With companies accountable to nothing, not even their statements to Congress and the White House, self-regulation is never going to work.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scoop-what-former-employees-of-openai-ad2f79-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
If internal and external AI governance isn't solved before the next big AI advance, we could be in serious trouble.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scoop-what-former-employees-of-openai-ab415f-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scoop-what-former-employees-of-openai-85114f
hedged language is not scored, even on a plain reading
2026-10-29
Assessment publishedBy this siteGary Marcus
So far, nothing in existing US law addresses the need for external governance under which companies cannot make decisions of potentially enormous magnitude on their own and external advisory bodies have a strong say in weighing risks against benefits.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scoop-what-former-employees-of-openai-7b2aee-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scoop-what-former-employees-of-openai-6498a7
The final-text check failed; overclaims: criteria.from_speaker lists "they could have proposed amendments months ago" and "including sun-setting the California law once sufficiently similar federal law existed" as the speaker's words. These are Saunders's draft-blog words that Marcus quotes. The drafter's own revision note says so..
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scoop-what-former-employees-of-openai-5da98c
The anthropic screen doubts the attribution.
2026-10-29
Assessment publishedBy this siteGary Marcus
The governance board Altman spoke of to The New Yorker in 2016 never materialized, and Marcus bets that if one does, it won't have any teeth unless a government mandate requires it to.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scoop-what-former-employees-of-openai-50f006-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
When and if some future form of AI brings new threats, we will be screwed, with our collective fates determined by a tiny number of individuals running a few companies with almost no accountability.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scoop-what-former-employees-of-openai-398424-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
OpenAI's stated justifications for opposing SB-1047, such as its argument that frontier AI regulation should be set at the federal level rather than through state laws, were meagre.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scoop-what-former-employees-of-openai-28bd91-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scoop-what-former-employees-of-openai-1e9885
The anthropic screen doubts the attribution.
2026-10-29
Assessment publishedBy this siteGary Marcus
ChatGPT is unreliable and untruthful: it makes arithmetic errors, fabricates biographical details, bungles word problems, and invents scientific phenomena.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-please-dont-let-your-chatbots-fae914-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
The worst thing about ChatGPT's wrong answer in the cited tweet is not that it is wrong but that it seems so convincing.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-please-dont-let-your-chatbots-d1ea68-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
ChatGPT does not contribute genuine ideas, design controlled experiments, or thoughtfully summarize literature.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-please-dont-let-your-chatbots-bde4b4-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
ChatGPT is a tool, closer to a spell-checker or stats package than to a scientist.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-please-dont-let-your-chatbots-ae8918-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The trend of listing ChatGPT as a co-author should not continue.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-please-dont-let-your-chatbots-690228-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Declaring ChatGPT a co-author says that you are more interested in being on trend than in being on target.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-please-dont-let-your-chatbots-4d9589-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Given its current scientific understanding, assigning authorship to ChatGPT disrespects the scientific process.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-please-dont-let-your-chatbots-3fc387-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
ChatGPT is clearly not sentient, and scientists should stop treating it as such because doing so misleads the public.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-please-dont-let-your-chatbots-372a23-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
ChatGPT should not be made a co-author, just as you wouldn't make your spell-checker or spreadsheet a co-author.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-please-dont-let-your-chatbots-337542-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Scientists should not co-author papers with ChatGPT.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-please-dont-let-your-chatbots-2e48f9-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
ChatGPT cannot reason well enough to make a substantive scientific contribution and has misunderstood physics, biology, psychology, and medicine.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-please-dont-let-your-chatbots-1e17cf-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scientists-please-dont-let-your-chatbots-0dd133
The anthropic screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
There should be immediate development of a global, neutral, non-profit International Agency for AI (IAAI), with guidance and buy-in from governments, large technology companies, non-profits, academia and society at large, aimed at collaboratively finding governance and technical solutions to promote safe, secure and peaceful AI technologies.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-governments-and-corporations-ed469b-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
We should worry about both near-term AI risks (such as LLM misinformation and misuse by bad actors) and long-term AI risks.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-governments-and-corporations-c55139-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
AI development needs to slow down and focus on research making AI systems trustworthy and reliable.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-governments-and-corporations-8ab7ed-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Neural networks do not understand the world.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scientists-governments-and-corporations-54ce6f-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Society should not leave AI's future entirely to industry self-regulation.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scarjo-is-just-the-tip-of-the-iceberg-fedea2-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The real issue is not whether OpenAI would win in court but what happens to all of us if a company with a track record for cutting ethical corners ends up first to AGI.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scarjo-is-just-the-tip-of-the-iceberg-6d06ee-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
OpenAI appears to be focused precisely on financial return and appears almost indifferent to some of the ways its product has already hurt large numbers of people, such as artists, writers and voiceover actors.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scarjo-is-just-the-tip-of-the-iceberg-5cbe4e-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scaling-is-over-the-bubble-may-be-eb414d
The final-text check failed: citations_correct.
2026-10-29
Assessment publishedBy this siteGary Marcus
Scaling of LLMs has run out as of April 2025.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scaling-is-over-the-bubble-may-be-d00cb4-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Sam Altman cannot be trusted.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scaling-is-over-the-bubble-may-be-c60d96-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The generative AI bubble might possibly be deflating as of April 2025.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scaling-is-over-the-bubble-may-be-aa7e98-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The AI financial bubble may be bursting as of April 2025.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scaling-is-over-the-bubble-may-be-a5c690-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
As of April 2025, all 25 of Marcus's predictions for 2025 seem to be on track.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scaling-is-over-the-bubble-may-be-8bffe6-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs still do not reason reliably as of April 2025.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scaling-is-over-the-bubble-may-be-719549-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Generative AI may be turning out to be a financial dud.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scaling-is-over-the-bubble-may-be-577d1a-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scaling-is-over-the-bubble-may-be-55e09f
The anthropic review does not accept e3 as contradicts.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs are not the right path for AI, and better alternative approaches are needed.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scaling-is-over-the-bubble-may-be-2b1773-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-scaling-hasnt-gotten-us-to-agi-or-bb88a3
The two reviews differ: anthropic historical_observation/not_applicable/supported, openai historical_observation/not_applicable/unresolved.
2026-10-29
Assessment publishedBy this siteGary Marcus
The AI field should rethink its approach, and that rethink is overdue.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-scaling-hasnt-gotten-us-to-agi-or-67c040-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus restates the point of his 2022 essay: AI probably needed new architectures beyond scaling LLMs.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-satya-nadella-and-the-three-stages-e09b95-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Resistance to criticism of scaling slowed scientific progress in AI.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-satya-nadella-and-the-three-stages-a08185-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Hundreds of billions of dollars went into LLM scaling on the assumption that scaling would keep working, and alternative approaches got little investment.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-satya-nadella-and-the-three-stages-71fb79-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Focusing on LLMs cost AI research almost half a decade. Investing in alternatives might have produced real progress toward robust, trustworthy AI.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-satya-nadella-and-the-three-stages-5a4431-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-satya-nadella-and-the-three-stages-33c6c1
The anthropic review reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
In the last couple of weeks before November 23, 2024, many other tech CEOs and others, including Marc Andreessen and Alexandr Wang, used the phrase "hitting a wall" and the idea of diminishing returns.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-satya-nadella-and-the-three-stages-1c0969-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
In recent weeks, as of November 2024, Marcus's 2022 conjecture has reached the stage of being regarded as self-evident, with Microsoft CEO Satya Nadella's remarks as an example. The conjecture is that LLM scaling laws are empirical observations that may not hold and that new approaches would be needed.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-satya-nadella-and-the-three-stages-0de1c0-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Nobody should ever trust Sam Altman.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altmans-pants-are-totally-on-ea1edb-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Sam Altman is now directly lying to the American public, not just to employees.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altmans-pants-are-totally-on-cd9c7c-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
The US government should not bail out OpenAI from its reckless spending; the idea is outrageous.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altmans-pants-are-totally-on-8cd9f8-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
In a podcast that was probably recorded in the last several days, Altman appears to have been laying the groundwork for loan guarantees.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altmans-pants-are-totally-on-8a025a-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altmans-pants-are-totally-on-7d9bf2
The anthropic review does not accept e1 as contradicts.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altmans-pants-are-totally-on-6fc4ca
The two reviews differ: anthropic historical_observation/not_applicable/mixed, openai historical_observation/not_applicable/unresolved.
2026-10-29
Assessment publishedBy this siteGary Marcus
Altman was launching a full court press for federal loan guarantees, likely in conjunction with Nvidia, which also seemed to be laying groundwork for a bailout.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altmans-pants-are-totally-on-3b935c-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The narrative about Sam Altman, which in November was that everybody loved him, has changed a lot.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altmans-fall-from-grace-de7b53-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Many people are reaching essentially the same perceptions of Sam Altman.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altmans-fall-from-grace-3561b0-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altmans-attitude-problem-f8e4c9
The anthropic review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
Sam Altman, despite outward appearances, is not a nice person.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altmans-attitude-problem-f72ce3-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Sam Altman's response at TED took no responsibility for what appears to be an obvious copyright infringement by an OpenAI system.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altmans-attitude-problem-e87680-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altmans-attitude-problem-d67204
The anthropic review reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus suspects that OpenAI's new Superintelligence report is just another Altman attempt to distract, in this case from the fact that the economics of OpenAI make no sense.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-unconstrained-by-the-truth-a6dbdf-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
As far as Marcus can tell, OpenAI's April 2026 Superintelligence report is not about anything real or imminent.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-unconstrained-by-the-truth-7a4936-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-unconstrained-by-the-truth-674386
The openai screen doubts the attribution.
2026-10-29
Assessment publishedBy this siteGary Marcus
OpenAI's economics make no sense.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-unconstrained-by-the-truth-3941ce-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
When Sam Altman is in trouble, he often tries to change the narrative with hype.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-unconstrained-by-the-truth-2959b2-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Sam Altman should not be trusted.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-unconstrained-by-the-truth-22ff23-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
As of April 2026, Sam Altman is in trouble, with people increasingly concerned about OpenAI's economics.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-unconstrained-by-the-truth-1533bc-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus says the presentation of o3 was problematic and gave no clean test of out-of-domain generalization, which suggests o3 may show the same distribution-shift problems as ever outside semi-closed domains where synthetic data can readily be generated.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-thinks-that-agi-is-basically-f2c303-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus does not share Sam Altman's confidence that OpenAI knows how to build AGI.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-thinks-that-agi-is-basically-db5a93-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus thinks several recent observations show how far AI is from robust intelligence.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-thinks-that-agi-is-basically-d11693-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-thinks-that-agi-is-basically-b66b53
The anthropic review does not accept e5 as supports.
2026-10-29
Assessment publishedBy this siteGary Marcus
In o1's best-performance cases there may have been heavy data augmentation that is not possible in more open-ended domains.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-thinks-that-agi-is-basically-ae7212-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
o1 is probably less robust than hoped, and some benchmark results may be fragile or hard to replicate.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-thinks-that-agi-is-basically-ad7a5f-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
OpenAI's o1 is losing money.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-thinks-that-agi-is-basically-784185-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-thinks-that-agi-is-basically-784185-2
The openai screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
Where data augmentation isn't possible, o1 may not be much better than GPT-4, for example on some language tasks.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-thinks-that-agi-is-basically-6c0fd9-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Test-time scaling is very expensive and will likely prove unreliable outside domains where data augmentation is adequate.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-thinks-that-agi-is-basically-199106-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-thinks-that-agi-is-basically-199106-2
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
So far Marcus has not seen a principled solution to even one of these longstanding problems, and he says that tells him a lot.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-thinks-that-agi-is-basically-16700a-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The lack of distinct, accessible, reliable database-style records leads to hallucinations, as Marcus noted in 2001.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-thinks-that-agi-is-basically-0a0fa4-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-then-and-now-dda869
The anthropic screen doubts the attribution.
2026-10-29
Assessment publishedBy this siteGary Marcus
Altman did not know how to build AGI in January 2025 and probably still does not as of February 2026.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-and-the-day-nvidias-meteoric-f9cb01-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-and-the-day-nvidias-meteoric-eb35c6
The anthropic review reads the claim differently.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-and-the-day-nvidias-meteoric-e6b92c
The final-text check failed: citations_correct.
2026-10-29
Assessment publishedBy this siteGary Marcus
Apart from Jensen Huang, nobody was more responsible for Nvidia's rise than Sam Altman, the showman who runs OpenAI and who repeatedly promised the moon.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-and-the-day-nvidias-meteoric-d1c794-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Nvidia's flat stock since August 2025 shows its rise has run out of gas, though it has not crashed.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-and-the-day-nvidias-meteoric-cf329c-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The market may now be ready, and desperate, for AI approaches beyond LLM hype.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-and-the-day-nvidias-meteoric-b7f6bc-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
AI tech stocks such as Nvidia, CoreWeave and Oracle, and OpenAI's reputation, will probably fall further.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-and-the-day-nvidias-meteoric-a99089-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-and-the-day-nvidias-meteoric-a99089-2
The openai screen reads the claim differently.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-and-the-day-nvidias-meteoric-a536b5
The anthropic review does not accept e5 as contradicts.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-and-the-day-nvidias-meteoric-91e28b
The openai screen reads the claim differently.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-and-the-day-nvidias-meteoric-8b0136
The two reviews differ: anthropic historical_observation/not_applicable/mixed, openai historical_observation/not_applicable/unresolved.
2026-10-29
Assessment publishedBy this siteGary Marcus
Altman's January 2025 claim that OpenAI was confident it knew how to build AGI was a bluff.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-and-the-day-nvidias-meteoric-7de474-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Circular financing in AI is a warning sign that the market is propped up rather than healthy.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-and-the-day-nvidias-meteoric-5a1c34-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-and-the-day-nvidias-meteoric-224614
The anthropic screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
With LLM mania subsiding, there might actually be a chance for newcomers to try new things.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-sam-altman-and-the-day-nvidias-meteoric-1671c1-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-and-the-day-nvidias-meteoric-159d0e
The anthropic screen reads the claim differently.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-sam-altman-and-copyright-then-and-c378c1
The anthropic screen reads the claim differently.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-rumors-of-agis-arrival-have-been-f61a0b
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
By original AGI standards, current AI systems remain limited.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-c64d40-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
AGI has not already been achieved; rumors that it has are greatly exaggerated.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-c33bcf-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Recent claims of achieving AGI conflate sophisticated statistical approximation with intelligence itself.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-c2d0d9-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-rumors-of-agis-arrival-have-been-b7f930
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
Claims that AGI has already arrived confuse high performance on benchmarks, which are often gameable, with real-world flexibility, which historically has been at the center of what it means to be an AGI.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-a0d93c-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Impressive task-level performance is not sufficient evidence of general intelligence.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-9921c1-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-rumors-of-agis-arrival-have-been-989eab
The openai review found the criteria unfair to the speaker's words.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-rumors-of-agis-arrival-have-been-7479db
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
Overestimating AI's cognitive capacities as it is embedded in decision-making risks misallocating trust, responsibility, and authority.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-65125a-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
In the cited direct comparison of human and LLM judgments of news source reliability, language models tended to issue confident classifications in the same cases where humans withheld or downgraded confidence, even when uncertainty was explicitly signaled in the input.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-62eb9c-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Chen et al.'s (2026) claims that AGI has been achieved depend on redefining the historical meaning of AGI.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-571185-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Current AI systems are powerful and increasingly useful tools, but they do not exhibit the flexible, self-directed competence that the original concept of AGI was intended to capture.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-520739-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Definitions that treat AGI as strong performance on standardized benchmarks miss the generality that was central to the original definitions, and so reflect a fundamental misunderstanding of how intelligence should be evaluated.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-501179-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-rumors-of-agis-arrival-have-been-4dd675
The anthropic screen finds the claim checkable, so it must be graded.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLM proofs of isolated open math problems, including Erdős problems, do not show general intelligence.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-197ebe-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
AI benchmarks are often easily gamed.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-0b1c6b-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Confusing increasingly sophisticated statistical approximation with general intelligence is not only a conceptual error but also a strategic misjudgment.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-090c74-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-rumors-of-agis-arrival-have-been-0779c8
The final-text check failed; overclaims: On the most-cited leaderboard, rank gains partly reflected benchmark-specific overfitting, not general quality..
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs' behavioral similarity to humans alone gives no insight into whether they have judgment, error correction, or reliable generalization.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-06f36e-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Across domains involving causality, value trade-offs, justification, and error monitoring, humans attend to reasons, stakes, and responsibility, whereas language models optimize for linguistic plausibility.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-rumors-of-agis-arrival-have-been-01ca07-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs are mimics and what they say is not always true.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-richard-dawkins-and-the-claude-delusion-ef103d-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Claude being able to draw on its training data to wax poetic about orgasm, which Marcus is sure it can do, does not mean Claude has ever felt one.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-richard-dawkins-and-the-claude-delusion-d4f6fc-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
His 2022 critique of LaMDA applies fully to Claude: it matches statistical patterns, its words lack meaning, it has no coherent understanding of the world, and it is not sentient.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-richard-dawkins-and-the-claude-delusion-d2d74d-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-richard-dawkins-and-the-claude-delusion-b7b0d5
The final-text check failed: citations_correct.
2026-10-29
Assessment publishedBy this siteGary Marcus
Claude is akin to a counterfeit person.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-richard-dawkins-and-the-claude-delusion-ad8890-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Machine intelligence, as in a chess computer, does not entail consciousness.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-richard-dawkins-and-the-claude-delusion-a11ee7-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
There is no reason to think Claude feels anything at all.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-richard-dawkins-and-the-claude-delusion-7e3f60-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
It does not follow from LLMs saying things people say that LLMs are like people.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-richard-dawkins-and-the-claude-delusion-688f63-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLM mimicry of human speech, however rich, proves very little about consciousness.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-richard-dawkins-and-the-claude-delusion-5b607f-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Claude's outputs come from a form of mimicry and are not reports of genuine internal states.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-richard-dawkins-and-the-claude-delusion-43e78e-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Dawkins should never have glorified Claude, which Marcus describes as akin to a counterfeit person.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-richard-dawkins-and-the-claude-delusion-136625-a1.