Changes, page 36 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 36 of 37: changes 7001 to 7200 of 7383, newest first. Dates are when the change was recorded.

2026-10-29
Assessment publishedBy this siteGary Marcus
No serious person thinks that AGI (artificial general intelligence) has been reached yet.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-e4aeeb-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Every claim he has seen that AGI has already been reached depends on lowering the definition of AGI.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-d1ad99-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Broad but shallow AI can never be guaranteed to be safe, human-compatible, or to do what we ask.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-b96150-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs remain superficial: their responses invariably have the right kind of language for any given problem, but they are rarely trustworthy.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-9e6c99-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Because LLMs remain unpredictable, humans should stay in the loop in most cases.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-96d618-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus proposes that LLMs should be called 'broad, shallow intelligence'.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-823a1f-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
No system has been shown to be robust across all the variations of the posted LLM failure examples under discussion (such as those from Mahabal and Fraser), though which specific examples fail varies across replications, perhaps in response to publicly shared examples.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-7f12a6-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
AI's unreliability will be reliably solved only once systems reason over concepts with deep understanding.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-6b7bd3-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Relying on shallow, unreliable AI could cause huge problems.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-6165fd-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agi-versus-broad-shallow-intelligence-559075
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
Humanity should make it a major research priority to move beyond broad, shallow AI toward safer, more reliable AI.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-55042d-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The depth LLMs seem to have at times, from regurgitating or replicating deep analyses found on the internet, is an illusion; genuine depth is lacking.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-4c62a6-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agi-versus-broad-shallow-intelligence-4550df
The two reviews differ: anthropic historical_observation/not_applicable/supported, openai historical_observation/not_applicable/unresolved.
2026-10-29
Assessment publishedBy this siteGary Marcus
Anyone claiming AGI has been achieved is either confused or hyping something.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-2a65f4-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
AI systems should solve problems by reasoning over concepts, not by surface analogy to text scraped from the internet.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-versus-broad-shallow-intelligence-0e0269-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agi-versus-broad-shallow-intelligence-040042
The openai review does not accept e2 as supports.
2026-10-29
Assessment publishedBy this siteGary Marcus
It is quite possible that no model worthy of the GPT-5 name will appear in 2025.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-isnt-coming-in-2025-and-gpt-5-fda39b-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agi-isnt-coming-in-2025-and-gpt-5-f81a5c
The anthropic screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
Musk's promise of 500,000 humanoid robots in three years seems very unlikely to be fulfilled.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-isnt-coming-in-2025-and-gpt-5-f7ab10-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
GPT-5 may well not be released in 2025, owing to data bottlenecks and inherent limits of LLMs.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-isnt-coming-in-2025-and-gpt-5-eb42ff-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The media should hold those who make unrealistic AI promises to account.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-isnt-coming-in-2025-and-gpt-5-d08aaf-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agi-isnt-coming-in-2025-and-gpt-5-cc7d8f
only a conditional forecast records a condition status
2026-10-29
Assessment publishedBy this siteGary Marcus
o3 is a real advance beyond GPT-4 in some domains, but it is likely not the quantum leap people would expect of the name GPT-5, and at a rumored $1000 per query it is certainly not at the price point anyone might have hoped for.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-isnt-coming-in-2025-and-gpt-5-bd27bd-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agi-isnt-coming-in-2025-and-gpt-5-5f0507
The two reviews differ: anthropic forecast_outcome/elapsed/supported, openai forecast_outcome/elapsed/unresolved.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus's hypothesis, which he has been floating since 2022, that scaling data alone would not be enough for robust AI has become increasingly prevalent.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-isnt-coming-in-2025-and-gpt-5-439e22-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
OpenAI is struggling to build a model worthy of the GPT-5 name.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-isnt-coming-in-2025-and-gpt-5-3af826-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
As of January 2025, GPT-5 has not been released despite nearly two years of claims of exponential progress.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-isnt-coming-in-2025-and-gpt-5-1add49-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agi-by-2027-ecfc15
The openai review does not accept e3 as supports.
2026-10-29
Assessment publishedBy this siteGary Marcus
It is imperative that we do better at reining in Silicon Valley than we have so far.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-by-2027-e9d3bc-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agi-by-2027-c93b79
The anthropic screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
Aschenbrenner's graph assumes all continued progress will be the same, ignoring qualitatively unsolved problems such as hallucinations, planning and reasoning, the possibility that synthetic data may not be as useful as natural data, and limits on energy distribution and compute.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-by-2027-b5cc78-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agi-by-2027-8e5807
The openai review found the criteria unfair to the speaker's words.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agi-by-2027-82f7ed
The openai review does not accept e2 as supports.
2026-10-29
Assessment publishedBy this siteGary Marcus
The double Y-axis on Aschenbrenner's graph makes no sense and presupposes its own conclusion.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-by-2027-6c1b48-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agi-by-2027-6476b0
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
Compared with a smart high schooler, GPT-4 can do some things better and some things worse, and it can do very few things reliably.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-by-2027-341718-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
We are woefully underprepared for AGI, whenever it comes, as Eliezer Yudkowsky has rightly argued for years.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-by-2027-32bbc1-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
We should be worried about AGI preparedness regardless of how long until AGI arrives.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-agi-by-2027-1c69dd-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agents-and-roi-faa534
The anthropic screen finds the claim checkable, so it must be graded.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-agents-and-roi-f0a458
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
If enough other big customers are disappointed with their AI returns the way Accenture was, "the whole jig" (an unspecified current AI arrangement) will be up.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-accenture-then-and-now-and-how-it-e64003-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-accenture-then-and-now-and-how-it-e57c41
The anthropic review does not accept e4 as contradicts.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-accenture-then-and-now-and-how-it-cd91b0
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
Claude Code will increase coders' productivity, unless technical debt cancels out the gains.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-accenture-then-and-now-and-how-it-c79d4e-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Overall, generative AI has not been delivering large gains, as many studies (MIT, McKinsey, Bain) show.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-accenture-then-and-now-and-how-it-a4ae66-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-accenture-then-and-now-and-how-it-46110f
The openai review found the criteria unfair to the speaker's words.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-accenture-then-and-now-and-how-it-3a1dd1
a observation claim cannot be assessed on the causal_assessment axis
2026-10-29
Assessment publishedBy this siteGary Marcus
Tokenmaxxing (maximizing AI token use) is in decline.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-accenture-then-and-now-and-how-it-3963ab-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-absolutely-amazing-ant-video-augurs-c1e501
The openai review reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
YouTube is likely to be ruined as a major educational resource almost overnight by AI-generated fake videos.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-absolutely-amazing-ant-video-augurs-b119d4-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-absolutely-amazing-ant-video-augurs-9e3fe2
The final-text check failed: rationale_supported_by_evidence; overclaims: Both reports concern Sora 2, a later version than the Sora model in Marcus's post..
2026-10-29
Assessment publishedBy this siteGary Marcus
Society faces turbulent times from AI-generated video.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-absolutely-amazing-ant-video-augurs-642d6e-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
A Sora-generated ant video got the ants' anatomy wrong: it did not show the six legs real ants have.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-absolutely-amazing-ant-video-augurs-4f6693-a3.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-absolutely-amazing-ant-video-augurs-1c8acc
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
Open-source counterparts to Sora will emerge.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-absolutely-amazing-ant-video-augurs-1c8acc-2-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Soon, a whole generation of children will be educated by AI-generated fake videos that look plausible to naive viewers but are factually incorrect.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-absolutely-amazing-ant-video-augurs-059b4e-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Autonomous vehicles still are not general enough that you can just plop down a car tuned in pilot studies in Menlo Park, SF and Arizona and expect it to drive gracefully and safely in Sicily or Mumbai.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-openai-breakthrough-f815d3-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-about-that-openai-breakthrough-e34ad1
The two reviews differ: anthropic historical_observation/not_applicable/supported, openai historical_observation/not_applicable/unresolved.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-about-that-openai-breakthrough-e28624
The two reviews differ: anthropic historical_observation/not_applicable/mixed, openai historical_observation/not_applicable/unresolved.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus is pretty skeptical about OpenAI's reported Q* breakthrough.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-openai-breakthrough-d87760-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
At least so far, OpenAI's Q* is not among the problems Marcus is worried about.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-openai-breakthrough-d48fe5-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Whether Q* pans out is still an empirical question, and Marcus himself does not yet know enough about the details of Q* to judge it with certainty.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-openai-breakthrough-9ffcc2-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus does not think Bing is already powered by OpenAI's alleged Q* breakthrough.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-openai-breakthrough-92337e-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
It is possible that OpenAI has a breakthrough that fundamentally changes the world.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-openai-breakthrough-3342bd-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
It is not realistic that OpenAI's Q* technique has already entirely changed the world within a few months.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-openai-breakthrough-2b1cb4-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
OpenAI's 2019 Rubik's cube robot result, achieved on a sensor-instrumented cube in controlled lab conditions, did not guarantee generalization to the open-ended real world.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-openai-breakthrough-284d5e-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
True level 5 driverless cars are probably still many breakthroughs away.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-openai-breakthrough-1a441b-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Claimed AI breakthroughs rarely prove general enough to live up to initial expectations, often working only in some contexts.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-openai-breakthrough-009872-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-about-that-matt-shumer-post-that-ea6da6
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 picture Shumer presents isn't realistic, however much people might wish it were true.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-matt-shumer-post-that-ea33b5-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-about-that-matt-shumer-post-that-da51c2
The final-text check failed: rationale_supported_by_evidence; overclaims: The gain was limited to the 50% success level..
2026-10-29
Assessment publishedBy this siteGary Marcus
People should not have taken Shumer's viral post about AI seriously.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-matt-shumer-post-that-d84a75-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs write code very fast, but some coders report burnout and only modest gains compared with that burnout.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-matt-shumer-post-that-89d9ae-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-about-that-matt-shumer-post-that-4cf74f
The openai screen reads the claim differently.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-about-that-matt-shumer-post-that-204d58
The two reviews differ: anthropic historical_observation/not_applicable/supported, openai historical_observation/not_applicable/unresolved.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-about-that-matt-shumer-post-that-19ff78
The two reviews differ: anthropic historical_observation/not_applicable/supported, openai historical_observation/not_applicable/unresolved.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-about-that-matt-shumer-post-that-1636a3
The final-text check failed: rationale_supported_by_evidence; overclaims: Trustworthiness beyond security, such as correctness and maintainability, is not measured directly..
2026-10-29
Assessment publishedBy this siteGary Marcus
The security of AI-generated code is seriously in doubt.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-about-that-matt-shumer-post-that-0c22c6-a3.
2026-10-29
Assessment publishedBy this siteGary Marcus
Building ever larger LLMs has never succeeded in reaching AGI.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-cf80f5-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
None of what Sutskever said should actually come as a surprise, especially not to readers of Marcus's Substack or to anyone who has followed Marcus over the years.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-ceb210-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-trillion-dollars-is-a-terrible-c72e95
The openai screen reads the claim differently.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-trillion-dollars-is-a-terrible-c04066
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
The AI investment-driven economy looks incredibly fragile.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-ae1240-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Trillion-dollar investments in ever larger experiments aimed at AGI may be delusional.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-9aef3a-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-trillion-dollars-is-a-terrible-9a927b
The anthropic review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
No LLM scaling experiment has solved the core problems of hallucination, generalization, planning and reasoning.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-9a927b-2-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-trillion-dollars-is-a-terrible-934689
The final-text check failed: citations_correct.
2026-10-29
Assessment publishedBy this siteGary Marcus
Most of the reasons to doubt LLMs have been known, though not widely accepted, for a very long time, and it all could have been avoided.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-74026d-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The AI boom just might end up taking down the economy.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-700bcf-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The AI industry may have wasted a trillion dollars and several years rediscovering what cognitive science already knew.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-6d4dd8-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Venture capitalists may continue to support LLM mania, at least for a while.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-6389a6-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
If the fallout spreads wider, the cost of the LLM bet could be much more than a trillion dollars.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-6189df-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
In the worst case, the impact of a deflated AI bubble could be immense.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-51e54f-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-trillion-dollars-is-a-terrible-3340e3
The anthropic screen reads the claim differently.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-trillion-dollars-is-a-terrible-2ef80f
The two reviews differ: anthropic historical_observation/not_applicable/mixed, openai historical_observation/not_applicable/unresolved.
2026-10-29
Assessment publishedBy this siteGary Marcus
Supposing for the sake of argument that LLMs have run their course and new ideas are needed, Marcus estimates, back of the envelope, the money cost of the field taking so long to recognize this at roughly a trillion dollars, much of it on Nvidia chips and massive salaries.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-1b7d29-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
A trillion dollars or more of LLM investment might be lost, with considerable economic and other collateral damage to society.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-160801-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-trillion-dollars-is-a-terrible-160801-2
The two reviews differ: anthropic causal_assessment/not_applicable/unresolved, openai causal_assessment/not_applicable/supported.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-trillion-dollars-is-a-terrible-0c2b43
The anthropic screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
Research papers (Apple's reasoning paper and the chain-of-thought mirage paper) demonstrated LLM reasoning and generalization failures.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-trillion-dollars-is-a-terrible-01726b-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-temporary-pause-on-training-extra-ebc7ba
The anthropic screen reads the claim differently.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-temporary-pause-on-training-extra-a62fb3
The anthropic review does not accept e2 as contradicts.
2026-10-29
Assessment publishedBy this siteGary Marcus
AI development should proceed with caution until the risks and benefits are better understood.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-temporary-pause-on-training-extra-6161b8-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-tale-of-two-liars-d05142
The anthropic review does not accept e1 as contradicts.
2026-10-29
Assessment publishedBy this siteGary Marcus
GPT cannot be trusted.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-tale-of-two-liars-8c6c20-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
GPT produces many falsehoods, but they are not really lies because GPT has no intent to deceive.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-tale-of-two-liars-8b951a-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
GPT autocompletes sentences; some outputs are true because training data is vast and some are false because training data is not infinite.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-tale-of-two-liars-356862-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-tale-of-two-liars-2cf3ba
a observation claim cannot be assessed on the causal_assessment axis
2026-10-29
Assessment publishedBy this siteGary Marcus
David Sacks should hold to the position he stated in early November that AI companies should be left to their own devices without government bailouts.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-tale-of-two-ai-capitalisms-e475d8-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
A tweet by White House AI Czar David Sacks appeared to lay the groundwork for a government bailout of overextended AI companies.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-tale-of-two-ai-capitalisms-965005-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-tale-of-two-ai-capitalisms-08e72e
The two reviews differ: anthropic forecast_outcome/open_horizon/not_adjudicable, openai forecast_outcome/open_horizon/unresolved.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-spate-of-outages-including-incidents-ae897a
The two reviews differ: anthropic historical_observation/not_applicable/mixed, openai historical_observation/not_applicable/unresolved.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-spate-of-outages-including-incidents-a40a6b
The openai review reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
AI may well come to write most code.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-spate-of-outages-including-incidents-9c31d2-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-spate-of-outages-including-incidents-9c31d2-2
The final-text check failed: rationale_supported_by_evidence; overclaims: Both items come from 2026, only months after the post. | That is the baseline the forecast starts from, but the same paper reports fast improvement..
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-spate-of-outages-including-incidents-520def
The final-text check failed: rationale_supported_by_evidence; overclaims: The 60% to 80% cost figure comes through a 2026 preprint's summary, not the original studies (Brooks; Lehman). | For AI agents, long-term maintenance proved harder than one-shot code changes..
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-spate-of-outages-including-incidents-3abe98
hedged language is not scored, even on a plain reading
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-sad-day-for-america-c9756e
The two reviews differ: anthropic historical_observation/not_applicable/contradicted, openai historical_observation/not_applicable/unresolved.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-sad-day-for-america-c55958
The anthropic screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
The US should adopt AI regulation between over- and under-regulation that fosters innovation while protecting citizens.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-sad-day-for-america-964678-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-sad-day-for-america-3875e2
The anthropic review reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
President Trump has trampled over states' rights and left US citizens in the lurch.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-sad-day-for-america-28b4ba-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
If an AI-related economic debacle occurs during Trump's term, he will be closely tied to it because of his closeness to Silicon Valley.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-sad-day-for-america-0ae9a8-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
People who live in the US, don't like what's happening, and don't want to leave American citizens almost completely vulnerable to the downsides of AI should contact their governor and state attorney general and let them know how they feel.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-sad-day-for-america-01c503-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus's hunch is that the AI preemption executive order will badly backfire politically on Trump, or at least on J.D. Vance.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-sad-day-for-america-00a173-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-new-ai-scaling-law-shell-game-fd7eda
The final-text check failed; overclaims: e3 explanation: "Contemporary inference-scaling research included explicit predictive models, so the entire field cannot be characterized as directionless experimentation." This rebuts a stronger claim than Marcus made. He said the new version is "much more 'let's try it and see what happens'", a comparison about the reframed scaling notion. He did not call the entire field directionless, and "directionless" is loaded wording..
2026-10-29
Assessment publishedBy this siteGary Marcus
Scaling alone has limited ability to advance AI.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-f63559-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-new-ai-scaling-law-shell-game-cd23c8
The anthropic review does not accept e3 as supports.
2026-10-29
Assessment publishedBy this siteGary Marcus
The new scaling laws (post-training and inference-time) will eventually stop holding.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-baa1e0-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
People should be financially wary of the new claims about multiple AI scaling laws.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-a1578e-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
OpenAI's own o1 research shows that scaling test-time compute helps on some problems but not others.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-9b4ced-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus guesses that o1's test-time scaling works best in closed domains where a lot of reliable synthetic data can be created, such as math problems, and not so well in open-ended domains.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-8de8be-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The AI field must move beyond scaling and pursue genuinely new ideas.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-82c2b3-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Making LLMs, which are inherently blind to factuality, run longer rather than better is a futile way to rescue them.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-7ca964-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-new-ai-scaling-law-shell-game-72c4f1
e1 is dated 2024-11, which is not inside the period described (2024-09-12 to 2024-11-24); it can only be context
2026-10-29
Assessment publishedBy this siteGary Marcus
Scaling performance by training models longer is no longer really holding.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-69084e-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
In some cases o1 does not perform better than GPT-4.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-2dc649-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
As of November 2024, a major retreat from scaling claims in the AI field has already begun.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-2a70ab-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
GPT-4 was better than GPT-3 in almost every way, whereas o1 is a far less uniform improvement.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-264969-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Scaling laws aren't really laws anymore, if they ever were, and the meaning of the word 'scaling' has been greatly devalued.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-new-ai-scaling-law-shell-game-09b05c-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-new-ai-scaling-law-shell-game-067a27
e2 is dated 2024-11, which is not inside the period described (2024-11-01 to 2024-11-24); it can only be context
2026-10-29
Assessment publishedBy this siteGary Marcus
OpenClaw devotees should beware the robustness of AI-generated code even more than other coders, by what Marcus puts as '10x, if not 100x'.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-model-that-produces-code-which-e3b929-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-model-that-produces-code-which-e36713
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
Next-word prediction gets us a surprisingly long way in writing code, but less far in making sure that code is robust.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-model-that-produces-code-which-dd13cd-a3.
2026-10-29
Assessment publishedBy this siteGary Marcus
Coders, especially inexperienced vibe coders, should be wary of the robustness of AI-generated code.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-model-that-produces-code-which-19357f-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
If a billion-dollar AI system cannot be used to solve a problem (Tower of Hanoi) that Herb Simon solved with AI in 1957 and that first-semester AI students solve routinely, then the chances that models like Claude or o3 will reach AGI seem truly remote.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-f5944d-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs may succeed on easy test sets and mislead users into thinking they have a generalizable solution when they do not.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-f38eef-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Models like o3 hallucinate much more than attentive humans and struggle with reliable diagrams.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-ebeec8-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus sarcastically suggests that getting to 'alignment' or 'safety' without reliability is a long shot.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-e9ceed-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs writing Python code to supplement their weaknesses is not reliable.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-e0245c-a3.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus is not interested in an 'AGI' that can't do arithmetic, and he certainly would not want to entrust global infrastructure or the future of humanity to such a system.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-d9a534-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
AGI should not be about perfectly replicating a human but should be about combining the best of both worlds: human adaptiveness with computational brute force and reliability.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-d68d25-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Anybody who thinks LLMs are a direct route to the sort of AGI that could fundamentally transform society for the good is kidding themselves.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-d22342-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs cannot reliably solve the Tower of Hanoi.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-c79e0c-a3.
2026-10-29
Assessment publishedBy this siteGary Marcus
Apple's new paper is pretty devastating to LLMs and is a powerful followup to a paper from many of the same authors last year.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-c521f3-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Other forms of deep learning, especially those that work well with symbols, may eventually thrive.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-98bf84-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
We are not going to 'extract the light cone' of the earth or 'solve physics' (Altman claims whose meaning Marcus questions) with systems that can't play Tower of Hanoi on a tower of 8 discs.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-88ba27-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs can't play chess, fold proteins, or run databases as well as conventional or special-purpose systems.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-753c69-a3.
2026-10-29
Assessment publishedBy this siteGary Marcus
Neural networks of various kinds can generalize within the training distribution of data they are exposed to, but their generalizations tend to break down outside that distribution.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-68a208-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
The LLM approach has limits that are becoming clearer every day.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-5c645c-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Claude gets under 80% accuracy on 7-disc Tower of Hanoi and essentially cannot solve 8 discs.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-3e2ec8-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-knockout-blow-for-llms-36d91f
The anthropic screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
o3 or Claude cannot simply be dropped into complex problems and expected to work reliably.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-27f761-a3.
2026-10-29
Assessment publishedBy this siteGary Marcus
For at least the next decade LLMs will continue to have uses, especially for coding, brainstorming and writing.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-25df94-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
LLMs are no substitute for good well-specified conventional algorithms.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-knockout-blow-for-llms-055f7c-a3.
2026-10-29
Assessment publishedBy this siteGary Marcus
High school students will love Galactica and will use it to fool and intimidate some of their teachers.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-few-words-about-bullshit-eaffe0-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-few-words-about-bullshit-e0cacd
The openai screen reads the claim differently.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-few-words-about-bullshit-949528
The anthropic screen doubts the attribution.
2026-10-29
Assessment publishedBy this siteGary Marcus
The Galactica examples Andrew Sundstrom shared are pitch-perfect but utterly bogus imitations of science and math, presented as the real thing.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-few-words-about-bullshit-2dc0d5-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-few-words-about-bullshit-2db47c
The final-text check failed: citations_correct.
2026-10-29
Assessment publishedBy this siteGary Marcus
Like Galactica, every other large language model the author has seen prevaricates (fabricates content) a lot.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-few-words-about-bullshit-097fc5-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-few-dark-words-about-chatbots-and-f3fc0a
The openai screen reads the claim differently.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-few-dark-words-about-chatbots-and-90183c
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
By now there have been several reports of chatbot-associated deaths, both suicides and murders, enough that Wikipedia has started keeping track of them.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-few-dark-words-about-chatbots-and-807451-a3.
2026-10-29
Assessment publishedBy this siteGary Marcus
Chatbots apparently inducing delusions in humans is a growing problem.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-few-dark-words-about-chatbots-and-6ece91-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Society should especially worry about chatbots influencing children, as teen deaths have shown.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-few-dark-words-about-chatbots-and-6e138b-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-few-dark-words-about-chatbots-and-6528f0
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus is increasingly convinced that, although generative AI has its uses, on balance it is not a step forward for humanity.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-few-dark-words-about-chatbots-and-63ad54-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-few-dark-words-about-chatbots-and-2a24e1
The openai review found the criteria unfair to the speaker's words.
2026-10-29
Assessment publishedBy this siteGary Marcus
The problem of chatbot-associated deaths is getting worse.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-few-dark-words-about-chatbots-and-150b11-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-few-dark-words-about-chatbots-and-098686
The final-text check failed: rationale_supported_by_evidence; overclaims: All of this was reported before the end of April 2023..
2026-10-29
Assessment publishedBy this siteGary Marcus
Because everybody is following the same playbook of ever more massive data and servers fed into ever-larger large language models, nobody has a real technical advantage in the generative AI race.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-f0a653-a3.
2026-10-29
Assessment publishedBy this siteGary Marcus
For all intents and purposes, the US-China generative AI race is basically a tie, with both countries serving many of their own customers using largely similar products.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-e20407-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Trump has come to believe that generative AI is a winner-take-all race, probably because of self-interested whisperings from Silicon Valley investors who often lean on FOMO about China to manipulate the government.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-dae19c-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-deeply-implausible-premise-is-behind-c5c782
The final-text check failed; overclaims: With e3, it shows the run lasted about four and a half months before OpenAI took over. | OpenAI took #1 from Google in August 2025, about four and a half months after Gemini 2.5 Pro took it. | The same article traces this lead back to Claude 3.5 Sonnet in June 2024. | That is a lasting lead in coding use, but it is market share, not the capability lead the claim mainly concerns..
2026-10-29
Assessment publishedBy this siteGary Marcus
Marcus does not see any plausible scenario in which either the US or China outright wins the generative AI race to the exclusion of the other.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-aa5969-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-deeply-implausible-premise-is-behind-96f183
e1 is dated 2026-01-02, which is not inside the period described (2025-12 to 2025-12); it can only be context; e2 is dated 2026-01-15, which is not inside the period described (2025-12 to 2025-12); it can only be context
2026-10-29
Assessment publishedBy this siteGary Marcus
Over the next few years both the US and China will build a lot of generative AI infrastructure, quite possibly significantly more than they need.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-80d210-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-deeply-implausible-premise-is-behind-71deb6
The final-text check failed: rationale_supported_by_evidence; overclaims: The only market data are a mid-2025 snapshot of enterprise LLM usage from before the post..
2026-10-29
Assessment publishedBy this siteGary Marcus
Nothing China can realistically do, short of full out nuclear war, is going to stop companies like Google, Microsoft, and Amazon from serving generative AI on the cloud.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-70451f-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
If LLMs turn out to be a dud, or are replaced by smaller, more efficient systems that don't demand such immense infrastructure, it may be all the more the case that the real winner is whichever country doesn't overextend itself to the point of financial ruin.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-6c7b3b-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Generative AI models are commodities.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-54906a-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Building US AI policy around a fantasy that the US is somehow going to crush China in an LLM war, or that China will crush the US, is misguided.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-5160b9-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
Nothing the United States can realistically do, short of full out nuclear war, is going to stop Chinese companies from serving generative AI on China-based cloud infrastructure.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-1fad20-a2.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-deeply-implausible-premise-is-behind-19daa2
e1 is dated 2026-01-02, which is not inside the period described (2025-12 to 2025-12); it can only be context; e2 is dated 2025-11-07, which is not inside the period described (2025-12 to 2025-12); it can only be context; e4 is dated 2025-11, which is not inside the period described (2025-12 to 2025-12); it can only be context; e5 is dated 2026-01-15, which is not inside the period described (2025-12 to 2025-12); it can only be context
2026-10-29
Assessment publishedBy this siteGary Marcus
The US-China lead in generative AI will be slight and flip back and forth, with neither retaining a sizeable lead for long.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-19daa2-2-a2.
2026-10-29
Assessment publishedBy this siteGary Marcus
Especially because chips like GPUs depreciate quickly, it may be that the real winner is whichever country doesn't overextend itself to the point of financial ruin in a foolish effort to win a race that can't be won.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-deeply-implausible-premise-is-behind-06b5c1-a1.
2026-10-29
HeldBy this siteGary Marcus
Held record marcus-a-cern-for-ai-and-the-global-governance-fb739e
The openai screen reads the claim differently.
2026-10-29
Assessment publishedBy this siteGary Marcus
There is no way to guarantee LLMs won't be jailbroken, and current LLMs can be jailbroken quickly and automatically.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-cern-for-ai-and-the-global-governance-f505da-a1.
2026-10-29
Assessment publishedBy this siteGary Marcus
It is likely that at least one of the 60+ elections worldwide in 2024 will be swayed by deepfakes.
First assessment published after both reviews and the final-text check passed.
Assessment marcus-a-cern-for-ai-and-the-global-governance-ca62a7-a1.