Claims checked against evidence by Gary Marcus, page 3 of 15
Deadlines are judged as of 2026-10-29. About Gary Marcus and the sources covered.
Claims checked against evidence by Gary Marcus (1471)
Each of these claims was checked against dated evidence.
Page 3 of 15: claims 201 to 300 of 1471, ordered by the date first said, newest first. To find a claim by its words, use search.
Assessment: SupportedSaid
Observation
“Anthropic refused to back down.”
Our reading: Anthropic refused to back down in its dispute with the US government over military use of its AI.
No horizon Evidence as of
Limitation: This assesses the company's position during the stated period, not the eventual outcome of negotiations.
Assessment: UnresolvedSaid
Conditional forecast
“If anything really bad comes of pushing premature AI too hard and too fast into to the military, he will own it.”
Our reading: If pushing premature AI too hard and too fast into the military leads to something really bad, Trump will be held responsible.
No deadline stated Evidence as of
Only if: if something really bad comes of pushing premature AI too hard and too fast into the military
Assessment: UnresolvedSaid
Causal claim
“One thing that Noble doesn’t mention is why OpenAI has struggled to make a profit; in my eyes, there are three reasons, that I have emphasized here before: the product is unreliable (placing an upper bound on how much many people be willing to pay), it is expensive to operate, and they have no technical moat, which has led to price wars.”
Our reading: In Marcus's view, one reason OpenAI has struggled to make a profit is that its product is unreliable, which places an upper bound on how much many people are willing to pay for it.
No horizon Evidence as of
Limitation: General AI survey findings do not directly measure OpenAI purchasing behavior.
Assessment: UnresolvedSaid
Forecast
“But that is exactly where we are headed.”
Our reading: We are headed exactly toward a situation where one individual, a Cabinet official, decides at gunpoint on mass surveillance and AI-fueled weapons, possibly nuclear, without humans in the loop.
No deadline stated Evidence as of
Limitation: Public records cannot establish the contents of undisclosed contracts or classified implementation.
Source: America, and probably the world, stands on a precipice.
Assessment: UnresolvedSaid
Observation
“Hegsseth’s demand would in principle extend to apply Anthropic’s software to nuclear weapons without humans in the loop.”
Our reading: Hegseth's demand would in principle allow Anthropic's software to be applied to nuclear weapons without humans in the loop.
No horizon Evidence as of
Limitation: Rests on interpretation of 'lawful' and 'in principle'.
Source: America, and probably the world, stands on a precipice.
Assessment: UnresolvedSaid
Observation
“Generative AI is NOT remotely reliable enough to make life or death decisions of magnitude.”
Our reading: Generative AI is nowhere near reliable enough to make major life-or-death decisions.
No horizon Evidence as of
Limitation: The claim leaves both the application and acceptable risk unspecified.
Source: 30 seconds to midnight? 15?
Assessment: SupportedSaid
Observation
“Although there are some practical use cases, such as coding, it is an inherently unreliable technology.”
Our reading: Generative AI has some practical uses such as coding but is an inherently unreliable technology.
No horizon Evidence as of
Limitation: Coding utility is supported here by self-reports; the experiment does not establish gains across coding tasks [e1, e3].
Assessment: SupportedSaid
Causal claim
“It is leading some people into serious mental health issues.”
Our reading: Generative AI is leading some people into serious mental health issues.
No horizon Evidence as of
Limitation: One clinical case does not establish population prevalence or quantify causal risk [e1].
Assessment: UnresolvedSaid
Observation
“Turns out it wasn’t, as Shira Ovide just reported:”
Our reading: Generative AI was not in fact contributing half of US GDP growth, contrary to the White House AI advisor's claim.
No horizon Evidence as of
Limitation: This does not validate the advisor's half-of-growth assertion.
Assessment: UnresolvedSaid
Observation
“I don’t know that Generative AI was literally a scam, but the people selling it have tried to sell it as if it were tantamount to artificial general intelligence, when it’s not.”
Our reading: Generative AI vendors have marketed it as tantamount to AGI, but it is not AGI.
No horizon Evidence as of
Limitation: Altman's quotation comes from a third-party transcript and does not establish what all vendors promised [e1].
Assessment: UnresolvedSaid
Causal claim
“Generative AI has its uses, but it hardly surprising that most business have struggled to find return on their investment, given the massive reliability issues that continue to plaguage it.”
Our reading: Most businesses have struggled to get return on generative AI investment because of its ongoing massive reliability issues.
No horizon Evidence as of
Limitation: NANDA's interviews, conference survey and public-initiative review do not establish prevalence among all generative-AI investors [e1].
Assessment: SupportedSaid
Observation
“They often produce workslop.”
Our reading: Large language models often produce low-quality 'workslop'.
No horizon Evidence as of
Limitation: Self-reported experience is not an independently graded output sample.
Assessment: UnresolvedSaid
Observation
“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—structural fault lines between performance alignment and genuine evaluative competence (Quattrociocchi et al., 2025).”
Our reading: 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.
No horizon Evidence as of
Limitation: A model can be trained for correctness while still making unreliable judgments.
Source: Rumors of AGI’s arrival have been greatly exaggerated
Assessment: UnresolvedSaid
Observation
“Behavioral similarity alone therefore provides no insight into whether the underlying processes support core components of general intelligence, such as judgment, error correction, or reliable generalization.”
Our reading: LLMs' behavioral similarity to humans alone gives no insight into whether they have judgment, error correction, or reliable generalization.
No horizon Evidence as of
Limitation: The strongest relevant studies use additional tests beyond behavioral similarity.
Source: Rumors of AGI’s arrival have been greatly exaggerated
Assessment: SupportedSaid
Observation
“Moreover, benchmarks are often easily gamed.”
Our reading: AI benchmarks are often easily gamed.
No horizon Evidence as of
Limitation: 'Often' has no threshold. The evidence covers prominent benchmarks, not all benchmarks.
Source: Rumors of AGI’s arrival have been greatly exaggerated
Assessment: SupportedSaid
Observation
“Reports that language models have produced correct proofs for isolated open problems in mathematics, including specific Erdős problems, do not alter this assessment.”
Our reading: LLM proofs of isolated open math problems, including Erdős problems, do not show general intelligence.
No horizon Evidence as of
Limitation: The 'do not alter this assessment' part is an evaluative judgment.
Source: Rumors of AGI’s arrival have been greatly exaggerated
Assessment: SupportedSaid
Observation
“Current AI systems are powerful and increasingly useful tools, but they do not exhibit the flexible, self-directed competence that the original concept of artificial general intelligence was intended to capture.”
Our reading: 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.
No horizon Evidence as of
Limitation: Most of the weight rests on one benchmark, ARC-AGI-3. Its 100% mark means matching human efficiency, not only solving the games.
Source: Rumors of AGI’s arrival have been greatly exaggerated
Assessment: UnresolvedSaid
Observation
“We also show that recent claims (Chen et al., 2026) about putative success on AGI hinge on redefining what AGI has historically meant.”
Our reading: Chen et al.'s (2026) claims that AGI has been achieved depend on redefining the historical meaning of AGI.
No horizon Evidence as of
Limitation: The packet lacks the complete Chen et al. argument required by the contract.
Source: Rumors of AGI’s arrival have been greatly exaggerated
Assessment: UnresolvedSaid
Observation
“Language models, by contrast, tended to issue confident classifications in the same cases, even when uncertainty was explicitly signaled in the input.”
Our reading: 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.
No horizon Evidence as of
Limitation: The supplementary material and usable underlying data were unavailable through the reader.
Source: Rumors of AGI’s arrival have been greatly exaggerated
Assessment: SupportedSaid
Observation
“Task-level performance, even when impressive, is not sufficient evidence of general intelligence.”
Our reading: Impressive task-level performance is not sufficient evidence of general intelligence.
No horizon Evidence as of
Limitation: The empirical example concerns particular systems and tasks, not every model available in February 2026.
Source: Rumors of AGI’s arrival have been greatly exaggerated
Assessment: UnresolvedSaid
Observation
“Such claims confuse high performance on benchmarks (often gameable) with real world flexibility, which historically has been at the center of what it means to be an artificial general intelligence.”
Our reading: 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.
No horizon Evidence as of
Limitation: The university report provides a direct author quotation but is not the complete Nature argument.
Source: Rumors of AGI’s arrival have been greatly exaggerated
Assessment: UnresolvedSaid
Observation
“Rumors that humanity has already achieved artificial general intelligence (AGI) have been greatly exaggerated.”
Our reading: AGI has not already been achieved; rumors that it has are greatly exaggerated.
No horizon Evidence as of
Limitation: This is a historical assessment; later model improvements do not contradict the claim.
Source: Rumors of AGI’s arrival have been greatly exaggerated
Assessment: SupportedSaid
Observation
“By the standards articulated in the original definitions of artificial general intelligence—robustness across environments, reliable generalization under novelty, and autonomous goal-directed behavior—current AI systems remain limited.”
Our reading: By original AGI standards, current AI systems remain limited.
No horizon Evidence as of
Limitation: The decisive benchmark results came out about five weeks after the post.
Source: Rumors of AGI’s arrival have been greatly exaggerated
Assessment: UnresolvedSaid
Observation
“Scammers will be among the first to adopt these tools. And indeed they already have;”
Our reading: Scammers are among the first adopters of AI impersonation and deepfake tools and are already using them.
No horizon Evidence as of
Limitation: Actual use does not by itself establish comparative adoption order.
Source: We URGENTLY need a federal law forbidding AI from impersonating humans
Assessment: SupportedSaid
Observation
“1Some state laws are trying to address these issues, but as we know the Federal government is trying to undermine the states’ abilities to pass and enforce AI regulation.”
Our reading: The US federal government is trying to undermine states' ability to pass and enforce AI regulation.
No horizon Evidence as of
Limitation: This finding establishes an attempted restriction, not that state laws had already been invalidated or enforcement stopped.
Source: We URGENTLY need a federal law forbidding AI from impersonating humans
Assessment: SupportedSaid
Observation
“Well here’s a new one, you can have OpenClaw call people, and pretend to be human.”
Our reading: OpenClaw can be used to phone people while pretending to be human, a new security risk.
No horizon Evidence as of
Limitation: Requires installation, configuration and a telephony provider.
Source: We URGENTLY need a federal law forbidding AI from impersonating humans
Assessment: UnresolvedSaid
Observation
“Anybody’s appearance can (given sufficient data) be faked now, at any time, for almost nothing.”
Our reading: Given sufficient data, deepfake tools can now fake anybody's appearance at any time for almost no cost.
No horizon Evidence as of
Limitation: The scientific synthesis is broader than person-specific video impersonation.
Source: We URGENTLY need a federal law forbidding AI from impersonating humans
Assessment: UnresolvedSaid
Observation
“OpenAI is losing money every quarter; even by their own (optimistic) calculations, they won’t make a profit for years.”
Our reading: OpenAI loses money every quarter and, even by its own optimistic projections, will not be profitable for years.
No horizon Evidence as of
Limitation: The packet contains no comparable series of quarterly net results through October–December 2025.
Assessment: SupportedSaid
Observation
“Google and Anthropic have each largely caught up; various Chinese companies are closing in.”
Our reading: Google and Anthropic have largely caught up with OpenAI, and Chinese companies are closing in.
No horizon Evidence as of
Limitation: The result depends on interpreting the statement as a claim about model capabilities.
Assessment: UnresolvedSaid
Conditional forecast
“If the venture capital dries up, they are toast. Talent will make a beeline for competitors.”
Our reading: If venture capital dries up, OpenAI will collapse and its talent will flee to competitors.
No deadline stated Evidence as of
Only if: if venture capital dries up
Assessment: UnresolvedSaid
Forecast
“Or even how he eventually walked that back to 2026 (which also won’t happen).”
Our reading: Elon Musk's promise of a million Tesla robotaxis, pushed back to 2026, will not be fulfilled in 2026.
Not yet due Evidence as of
Limitation: The December 31, 2026 deadline had not passed at the October 28 evaluation date.
Source: Promises are cheap
Assessment: SupportedSaid
Observation
“Back on planet earth, according to December’s Remote Labor Index, only 2.5% of human “online tasks” (physical tasks were excluded) could be done by AI.”
Our reading: According to the December 2025 Remote Labor Index, AI could do only 2.5% of human online tasks.
No horizon Evidence as of
Limitation: The judgment depends on keeping the claim within the named benchmark’s scope.
Source: Promises are cheap
Assessment: SupportedSaid
Observation
“The security of the autogenerated code is very much in question, as for example this article last week noted (precisely as this substack has long warned):”
Our reading: The security of AI-generated code is seriously in doubt.
No horizon Evidence as of
Limitation: One main study comes from a security vendor with a commercial interest in highlighting risk. The other is a preprint.
Source: About that Matt Shumer post that has nearly 50 million views
Assessment: SupportedSaid
Observation
“LLMs write code wicked fast, but some coders are starting to report burnout, and only relatively modest gains relative to that burnout, as a new story from Connie Loizos at Tech Crunch illustrates”
Our reading: LLMs write code very fast, but some coders report burnout and only modest gains compared with that burnout.
No horizon Evidence as of
Limitation: The evidence is a small set of reports: one interviewed engineer and one anonymous forum commenter. It does not show how common burnout is among coders.
Source: About that Matt Shumer post that has nearly 50 million views
Assessment: SupportedSaid
Observation
“LLMs still hallucinate and continue to make boneheaded errors.”
Our reading: As of February 2026, LLMs still hallucinate and still make boneheaded errors.
No horizon Evidence as of
Limitation: Vectara measures faithfulness in summaries only; it does not cover other kinds of hallucination.
Source: BREAKING: LLM “reasoning” continues to be deeply flawed
Assessment: SupportedSaid
Observation
“And reasoning is still one of the core issues.”
Our reading: As of February 2026, reasoning is still one of the core unsolved problems for LLMs.
No horizon Evidence as of
Limitation: 'Core issue' is qualitative; no metric ranks reasoning failures against other limitations.
Source: BREAKING: LLM “reasoning” continues to be deeply flawed
Assessment: UnresolvedSaid
Observation
“Although there is some clear value in it, the costs are high, and relative to those costs, the revenue has been modest.”
Our reading: Generative AI has some clear value, but its costs are high and its revenue has been modest relative to those costs.
No horizon Evidence as of
Limitation: The financial reporting concerns one provider.
Source: Some disconcerting facts about AI and banking that may have profound consequences
Assessment: UnresolvedSaid
Forecast
“Let me add that I do think that some form of AI will eventually actually yield the returns people were anticipating for Generative AI.”
Our reading: Some form of AI will eventually yield the returns people anticipated from generative AI.
No deadline stated Evidence as of
Limitation: Goldman Sachs provides an illustrative forecast, not realized gains or a threshold Marcus explicitly adopted.
Source: Some disconcerting facts about AI and banking that may have profound consequences
Assessment: SupportedSaid
Observation
“Not an all out crash, but an unmistakable sign that its ascent has run out of gas.”
Our reading: Nvidia's flat stock since August 2025 shows its rise has run out of gas, though it has not crashed.
No horizon Evidence as of
Limitation: Prices are intraday figures from Barron's, not official closes.
Source: Sam Altman and the day Nvidia’s meteoric rise came to an end
Assessment: SupportedSaid
Observation
“They have also faced a lot of backlash for being reckless and irresponsible; hardly a good brand name.”
Our reading: xAI has faced a lot of backlash for being reckless and irresponsible, which is hardly a good brand name.
No horizon Evidence as of
Limitation: The brand comment is assessed qualitatively.
Assessment: UnresolvedSaid
Causal claim
“Me, I think the merger is really a kind of bailout, to give a lot of cash to a company that is otherwise in distress.”
Our reading: Marcus thinks the SpaceX-xAI merger is really a kind of bailout, meant to give a lot of cash to xAI, which is otherwise in distress.
No horizon Evidence as of
Limitation: The fundraising announcement does not disclose unrestricted liquidity after outstanding commitments [e2].
Assessment: SupportedSaid
Observation
“OpenClaw itself is basically a cascade of LLMs.”
Our reading: OpenClaw is basically a cascade of LLMs.
No horizon Evidence as of
Limitation: Cascade is informal terminology.
Source: OpenClaw (a.k.a. Moltbot) is everywhere all at once, and a disaster waiting to happen
Assessment: UnresolvedSaid
Causal claim
“The point is that when it costs pennies to coordinate a fake mob and moments to counterfeit a human identity, the public square is left wide open to attack.”
Our reading: When it costs pennies to coordinate a fake mob and takes moments to counterfeit a human identity, the public square is left wide open to attack.
No horizon Evidence as of
Only if: it costs pennies to coordinate a fake mob
Assessment: SupportedSaid
Observation
“Today’s swarms, now emerging, can coordinate fleets of synthetic personas—sometimes with persistent identities—and move in ways that are hard to distinguish from real communities.”
Our reading: Emerging AI bot swarms can coordinate fleets of synthetic personas that are hard to distinguish from real communities.
No horizon Evidence as of
Limitation: 'Hard to distinguish' has no measure. The evidence does not show how long platforms or ordinary users took to detect these networks.
Assessment: UnresolvedSaid
Observation
“Even so-called “thinking” AI models are vulnerable to this,”
Our reading: Even so-called 'thinking' (reasoning) AI models are vulnerable to training-data poisoning by swarms.
No horizon Evidence as of
Limitation: The evidence includes no audit that names reasoning models and measures how often they repeat narratives seeded by grooming.
Assessment: SupportedSaid
Observation
“Venture capital is already helping industrialize astroturfing: Doublespeed, backed by Andreessen Horowitz, advertises a way to “orchestrate actions on thousands of social accounts” and to mimic “natural user interaction” on physical devices so the activity appears human.”
Our reading: Venture capital is already helping industrialize astroturfing; for example, Andreessen Horowitz-backed Doublespeed advertises a way to orchestrate actions on thousands of social accounts and to mimic natural user interaction on physical devices so the activity appears human.
No horizon Evidence as of
Limitation: Historical advertising wording rests on accessible reporting because the primary website has changed and the original 404 Media report was only partially accessible.
Assessment: UnresolvedSaid
Conditional forecast
“If we require this kind of “proof-of-human” for high-reach interactions, we make it mathematically difficult and financially ruinous for one operator to secretly run ten thousand accounts.”
Our reading: If verified-yet-anonymous proof-of-human is required for high-reach interactions, it will make it mathematically difficult and financially ruinous for one operator to secretly run ten thousand accounts.
No deadline stated Evidence as of
Only if: this kind of proof-of-human is required for high-reach interactions
Assessment: SupportedSaid
Observation
“And it’s just getting worse.”
Our reading: The problem of chatbot-associated deaths is getting worse.
No horizon Evidence as of
Limitation: The assessment counts reported cases and lawsuits. It does not use measured death rates, and the evidence includes no per-user rate.
Assessment: SupportedSaid
Observation
“And then there is the growing problem of chatbots apparently inducing delusions in humans, that Kashmir Hill and others have been documenting.”
Our reading: Chatbots apparently inducing delusions in humans is a growing problem.
No horizon Evidence as of
Limitation: The evidence covers documented cases, not true incidence. Awareness may inflate the trend.
Assessment: SupportedSaid
Observation
“By now, there have been several reports of chatbot-associated deaths (both suicide and murder), enough that wikipedia has started keeping track.”
Our reading: By now there have been several reports of chatbot-associated deaths, both suicides and murders, enough that Wikipedia has started keeping track of them.
No horizon Evidence as of
Limitation: The reported associations do not establish that chatbots caused the deaths; the hearing account and complaint record allegations [e2, e3].
Assessment: SupportedSaid
Observation
“That deal was OpenAI’s to lose. And they lost it.”
Our reading: OpenAI lost the Apple Siri generative AI deal to Google.
No horizon Evidence as of
Limitation: Support depends on lost meaning did not secure the opportunity.
Assessment: ContradictedSaid
Observation
“LLMs are commodities; DeepSeek proved that.”
Our reading: LLMs are commodities, as DeepSeek demonstrated.
No horizon Evidence as of
Limitation: The demand study covers one marketplace through April 2025, not the whole market through January 13, 2026 [e2].
Assessment: UnresolvedSaid
Observation
“OpenAI has never made a profit.”
Our reading: OpenAI has never made a profit.
No horizon Evidence as of
Limitation: The financial evidence is news reporting about statements, not directly inspected consolidated statements.
Assessment: SupportedSaid
Observation
“Many, many people were disappointed with GPT-5, arguably among the most overhyped products in history.”
Our reading: Many, many people were disappointed with GPT-5, which was arguably among the most overhyped products in history.
No horizon Evidence as of
Limitation: Support applies to substantial disappointment, not the qualified historical ranking.
Assessment: SupportedSaid
Observation
“I warned that LLMs might plateau within a year or two, and they have, that OpenAI would be mired in a lot of litigation, and they are.”
Our reading: OpenAI is mired in a lot of litigation.
No horizon Evidence as of
Limitation: The evidence covers one consolidated litigation group, not OpenAI’s complete litigation portfolio.
Assessment: UnresolvedSaid
Observation
“Anthropic has started to eat a lot of OpenAI’s lunch, taking a lot of corporate customers away.”
Our reading: Anthropic is taking many corporate customers away from OpenAI.
No horizon Evidence as of
Limitation: Menlo has an investment in Anthropic.
Assessment: UnresolvedSaid
Observation
“Over the last couple years, Altman’s public credibility has steadily waned (see e.g., Karen Hao’s bestseller, and my earlier essay on Altman in The Guardian).”
Our reading: Altman's public credibility has steadily declined over the last couple of years.
No horizon Evidence as of
Limitation: Historical values in an interactive tracker were not available in the page-reader output.
Assessment: SupportedSaid
Observation
“Most importantly, other companies like Anthropic and Google caught up; xAI and various Chinese companies aren’t far behind.”
Our reading: Anthropic and Google have caught up with OpenAI, and xAI and Chinese companies are not far behind.
No horizon Evidence as of
Limitation: The result depends on the ordinary competitive-tier reading of caught up and not far behind.
Assessment: UnresolvedSaid
Observation
“Those price wars did in fact come to pass; users costs per token are roughly a hundred times lower two year later.”
Our reading: LLM price wars occurred and user costs per token are roughly 100 times lower than two years earlier.
No horizon Evidence as of
Limitation: A supported component is not enough to support the full conjunction.
Assessment: SupportedSaid
Observation
“I also warned that LLMs would continue to have a truthiness problem and they have.”
Our reading: LLMs have continued to have a truthfulness problem.
No horizon Evidence as of
Limitation: This establishes persistence, not a uniform error rate across models or tasks.
Assessment: UnresolvedSaid
Observation
“(Altman got famous hinting at the supposed miracles of GPT-5; by any reasonable measure it was both late and short of expectations.)”
Our reading: GPT-5 was both late and short of expectations.
No horizon Evidence as of
Limitation: The reported postponement concerns an approximate public timeline, not a precise missed deadline [e1].
Assessment: UnresolvedSaid
Causal claim
“DeepSeek launched a price war that forced Altman to cut prices dramatically.”
Our reading: DeepSeek's price war forced OpenAI to cut prices dramatically.
No horizon Evidence as of
Limitation: A company's public efficiency explanation may omit commercial considerations.
Assessment: SupportedSaid
Observation
“Update: This is consistent with the finding of the Remote Labor Index that AI could only do about 2.5% of jobs, reported recently by the Washington Post.”
Our reading: Per the Remote Labor Index, AI can do only about 2.5% of jobs, consistent with LLMs adding little quantifiable value.
No horizon Evidence as of
Limitation: The benchmark result does not establish the share of all occupations AI can perform or aggregate economic value from LLM use [e1].
Source: Let’s be honest, Generative AI isn’t going all that well
Assessment: UnresolvedSaid
Observation
“They still aren’t adding a lot of quantifiable value to the world:”
Our reading: LLMs still are not adding much quantifiable value to the world.
No horizon Evidence as of
Limitation: The claim lacks a threshold and an agreed measure of value.
Source: Let’s be honest, Generative AI isn’t going all that well
Assessment: SupportedSaid
Observation
“LLMs can still cannot be trusted:”
Our reading: LLMs still cannot be trusted.
No horizon Evidence as of
Limitation: The supplied AA-Omniscience abstract does not specify whether browsing was enabled [e1].
Source: Let’s be honest, Generative AI isn’t going all that well
Assessment: UnresolvedSaid
Observation
“A large fraction of what LLMs do is mostly just memorization (and Hinton was on the wrong side of this argument):”
Our reading: A large fraction of what LLMs do is mostly memorization, and Hinton was wrong on this point.
No horizon Evidence as of
Limitation: Large fraction has no stated numerical threshold.
Source: Let’s be honest, Generative AI isn’t going all that well
Assessment: UnresolvedSaid
Conditional forecast
“If LeCun pulls the same stuff in his new company, morale will be deathly low.”
Our reading: If LeCun keeps failing to credit others' work at his new company, morale there will be very low.
No deadline stated Evidence as of
Only if: if LeCun fails to acknowledge others' work in his new company
Assessment: UnresolvedSaid
Causal claim
“My own view, which I stand by, is that we need “neurosymbolic” contributions from both traditions if we are to get to trustworthy AI.”
Our reading: In Marcus's stated view, which he stands by, getting to trustworthy AI requires neurosymbolic contributions from both the neural network and classical symbolic traditions.
No horizon Evidence as of
Limitation: The evidence provides no agreed application scope, failure tolerance, or boundary for what counts as a contribution from either tradition.
Assessment: UnresolvedSaid
Observation
“Of course, the LLM companies have armies of humans; in my own brief experiments a couple hours later, OpenAI had apparently already patched this particular problem.”
Our reading: LLM companies have armies of humans, and when Marcus ran his own brief experiments a couple of hours later, OpenAI had apparently already patched this particular problem.
No horizon Evidence as of
Limitation: Private experiments may have occurred without a public record.
Assessment: SupportedSaid
Observation
“Why LLMs just can’t stop hallucinating.”
Our reading: LLMs are unable to stop hallucinating.
No horizon Evidence as of
Limitation: The Vectara figures come from one benchmark of summarization faithfulness. They may not reflect other tasks or other ways of measuring factual accuracy.
Source: AI Cartoon of The Year — and five (re)readings for 2026
Assessment: UnresolvedSaid
Forecast
“Neurosymbolic AI will displace pure LLMs over time; here’s why.”
Our reading: Neurosymbolic AI will eventually displace pure LLMs.
No deadline stated Evidence as of
Limitation: Neither 'neurosymbolic' nor 'pure LLM' is defined. Whether LLMs that use tools or agent harnesses count as neurosymbolic is disputed, and the outcome may depend on that choice.
Source: AI Cartoon of The Year — and five (re)readings for 2026
Assessment: UnresolvedSaid
Forecast
“And LLMs will continue to exist.”
Our reading: LLMs will continue to exist even after the AI bubble falls apart.
No deadline stated Evidence as of
Limitation: The packet does not establish when or whether the anticipated broad collapse occurred.
Assessment: UnresolvedSaid
Causal claim
“And without reliability, profits are limited.”
Our reading: Without reliability, profits from AI are limited.
No horizon Evidence as of
Limitation: The usable evidence is a reputable news report because the located primary documents could not be read with the required tool.
Assessment: UnresolvedSaid
Forecast
“But the economics don’t make sense, and never will, in large part because of the core technical problems that I have stressed repeatedly here and elsewhere since 2019.”
Our reading: The economics of LLMs will never make sense.
No deadline stated Evidence as of
Limitation: The detailed financial evidence concerns one major developer, not every LLM business.
Assessment: SupportedSaid
Observation
“The technical problems are not new.”
Our reading: The core technical problems of LLMs are not new.
No horizon Evidence as of
Limitation: The evidence establishes continuity in broad failure categories, not identical causes across model generations.
Assessment: UnresolvedSaid
Causal claim
“This cold reality in turns undermines a large fraction of the use cases that people initially fantasized about.”
Our reading: The inherent limitations of LLMs undermine a large fraction of the use cases people initially imagined.
No horizon Evidence as of
Limitation: The studies examine different models, periods, tasks, and forms of human oversight.
Assessment: UnresolvedSaid
Observation
“What is new that they have at last become widely recognized, not just as the transitory “bugs” the industry wanted you to think of, but as inherent limitations that flow from the very design of LLMs they really are.”
Our reading: LLMs' technical problems are now widely recognized as inherent limitations of their design, not transitory bugs.
No horizon Evidence as of
Limitation: A selection of technical papers is not a representative opinion survey.
Assessment: SupportedSaid
Observation
“This is in line with the July METR study showing that coding tools sometimes impair productivity rather than increasing it.”
Our reading: The July METR study showed that AI coding tools sometimes impair productivity rather than increase it.
No horizon Evidence as of
Limitation: The trial involved experienced developers working on familiar open-source projects.
Assessment: SupportedSaid
Observation
“But many others have found that LLMs are not quite as advertised.”
Our reading: Many others have found that LLMs are not quite as good as advertised.
No horizon Evidence as of
Limitation: The advertising comparison is qualitative and supplied by interpretation.
Assessment: UnresolvedSaid
Forecast
“Human domestic robots like Optimus and Figure will be all demo and very little product.”
Our reading: In 2026, humanoid domestic robots such as Optimus and Figure will be mostly demos with very little actual product.
Not yet due Evidence as of
Limitation: The evidence includes no count of Optimus, Figure or NEO units delivered to consumer homes in 2026.
Source: Six (or seven) predictions for AI 2026 from a Generative AI realist
Assessment: UnresolvedSaid
Forecast
“2025 will be known as the year of the peak bubble, and also the moment at which Wall Street began to lose confidence in generative AI.”
Our reading: 2025 will come to be seen as the peak of the generative AI bubble and when Wall Street began losing confidence in generative AI.
No deadline stated Evidence as of
Limitation: The cited commentary is a secondary summary; no primary record can by itself establish collective historical recognition.
Source: Six (or seven) predictions for AI 2026 from a Generative AI realist
Assessment: UnresolvedSaid
Observation
“Few AI companies aside from Nvidia are making a profit, and nobody has much of a technical moat.”
Our reading: Few AI companies other than Nvidia are profitable, and no AI company has much of a technical moat.
No horizon Evidence as of
Limitation: Company-level profits do not isolate profits from AI products.
Source: Six (or seven) predictions for AI 2026 from a Generative AI realist
Assessment: UnresolvedSaid
Forecast
“Work on new approaches such as world models and neurosymbolic will escalate.”
Our reading: In 2026, work on new approaches such as world models and neurosymbolic AI will escalate.
Not yet due Evidence as of
Limitation: This is a qualitative trend forecast with several possible measures.
Source: Six (or seven) predictions for AI 2026 from a Generative AI realist
Assessment: UnresolvedSaid
Forecast
“We won’t get to AGI in 2026 (or 7).”
Our reading: AGI will not be achieved in 2026 or 2027.
Not yet due Evidence as of
Limitation: The evidence covers only OpenAI's statements up to about late August 2026. It does not cover other labs or later events.
Source: Six (or seven) predictions for AI 2026 from a Generative AI realist
Assessment: UnresolvedSaid
Forecast
“there will be tons of lab demos but getting these robots to work in people’s homes will be very very hard, as Rodney Brooks has said many times.”
Our reading: There will be many lab demos of humanoid robots in 2026, but getting them to work in homes will be very hard.
Not yet due Evidence as of
Limitation: The evidence includes no independent 2026 home trials measuring task success, failures or human intervention.
Source: Six (or seven) predictions for AI 2026 from a Generative AI realist
Assessment: UnresolvedSaid
Forecast
“I don’t expect my predictions to be as on target this year as last, for a happy reason: across the field, the intellectual situation has gone from one that was stagnant (all LLMs all the time) and unrealistic (“AGI is nigh”) to one that is more fluid, more realistic, and more open-minded.”
Our reading: Marcus's 2026 predictions will be less accurate than his 2025 ones because the field has become more fluid, realistic, and open-minded.
Not yet due Evidence as of
Limitation: The annual prediction sets require an explicit, consistent scoring method.
Source: Six (or seven) predictions for AI 2026 from a Generative AI realist
Assessment: UnresolvedSaid
Forecast
“No country will take a decisive lead in the GenAI “race”.”
Our reading: No country will take a decisive lead in the generative AI race in 2026.
Not yet due Evidence as of
Limitation: 'Decisive lead' has no threshold, and lead could be measured on several axes.
Source: Six (or seven) predictions for AI 2026 from a Generative AI realist
Assessment: UnresolvedSaid
Forecast
“Right now, as it happens, China is at most a few months behind; I expect the lead to be slight and to flip back and forth, with nobody retaining a sizeable a lead for long.”
Our reading: The US-China lead in generative AI will be slight and flip back and forth, with neither retaining a sizeable lead for long.
No deadline stated Evidence as of
Limitation: The evidence comes from one capability index, the Epoch Capabilities Index.
Source: The core misconception that is driving American AI policy
Assessment: UnresolvedSaid
Forecast
“Conversely, 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.”
Our reading: 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.
No deadline stated Evidence as of
Limitation: Restrictions on access to advanced compute must not be equated with cessation of existing inference services.
Source: The core misconception that is driving American AI policy
Assessment: MixedSaid
Observation
“Coke and Pepsi are commodities; genAI models are, too.”
Our reading: Generative AI models are commodities.
No horizon Evidence as of
Limitation: The buyer evidence comes from a single survey of builders by a venture firm, from mid-2025, months before the December 2025 post.
Source: The core misconception that is driving American AI policy
Assessment: UnresolvedSaid
Forecast
“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.”
Our reading: 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.
No deadline stated Evidence as of
Limitation: The operational examples do not provide an exhaustive history of every provider.
Source: The core misconception that is driving American AI policy
Assessment: UnresolvedSaid
Forecast
“Instead, part of the actual outcome over the next few years will be that both countries build a lot of generative AI infrastructure — quite possibly significantly more than they actually need.”
Our reading: 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.
Not yet due Evidence as of
Limitation: The horizon is vague. The end date of 2030 is an interpretation.
Source: The core misconception that is driving American AI policy
Assessment: UnresolvedSaid
Forecast
“I literally don’t see any plausible scenario in which either nation outright “wins” the generative AI race to the exclusion of the other.”
Our reading: 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.
No deadline stated Evidence as of
Limitation: Service announcements do not constitute a comprehensive global market census.
Source: The core misconception that is driving American AI policy
Assessment: UnresolvedSaid
Observation
“For all intents and purposes, the race is basically a tie – with both countries serving many of their own customers, using largely similar products.”
Our reading: 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.
No horizon Evidence as of
Limitation: US survey evidence covers teenagers, not the entire domestic market.
Source: The core misconception that is driving American AI policy
Assessment: UnresolvedSaid
Causal claim
“Importantly, since everybody is following the same playbook—ever more massive data, ever more massive servers, fed into ever-larger large language models—nobody has a real technical advantage here.”
Our reading: 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.
No horizon Evidence as of
Limitation: DeepSeek’s findings are a developer self-assessment, without comparable measurements establishing the practical magnitude of the disclosed deficits [e1].
Source: The core misconception that is driving American AI policy
Assessment: UnresolvedSaid
Conditional forecast
“And because he has become so tight with Silicon Valley he will also be closely tied to any AI-tinged economic debacle that happens on his watch.”
Our reading: 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.
Not yet due Evidence as of
Only if: if an AI-tinged economic debacle happens on his watch
Source: A sad day for America
Assessment: SupportedSaid
Observation
“As anticipated here yesterday, Trump’s decision to allow Nvidia to sell H200s to China hasn’t gone over well.”
Our reading: Trump's decision to allow Nvidia to sell H200 chips to China has not been well received.
No horizon Evidence as of
Limitation: This assesses substantial backlash, not majority public opinion.
Source: Has the Chinese government figured out the fact that GPUs ≠ AGI?
Assessment: SupportedSaid
Causal claim
“Part of Beijing’s response is no doubt about protectionism.”
Our reading: China's reluctance to buy H200s is partly driven by protectionism favoring homegrown companies like Huawei.
No horizon Evidence as of
Limitation: The evidence supports the substance of the claim more strongly than the certainty conveyed by “no doubt.”
Source: Has the Chinese government figured out the fact that GPUs ≠ AGI?
Assessment: UnresolvedSaid
Observation
“As noted yesterday, the H200 move is hardly popular with either party.”
Our reading: The decision to allow H200 sales to China is unpopular with both US political parties.
No horizon Evidence as of
Limitation: Party membership and the required breadth of opposition are unspecified.
Source: Has the Chinese government figured out the fact that GPUs ≠ AGI?
Assessment: SupportedSaid
Observation
“Same thing with proposed preemption of states’ rights to regulate AI.”
Our reading: Proposed federal preemption of states' rights to regulate AI is unpopular with both US political parties.
No horizon Evidence as of
Limitation: The precise preemption proposal is not named.
Source: Has the Chinese government figured out the fact that GPUs ≠ AGI?