Claims that can’t be checked by Gary Marcus, page 13 of 24
Deadlines are judged as of 2026-10-29. About Gary Marcus and the sources covered.
Claims that can’t be checked by Gary Marcus (2353)
Public evidence cannot settle these claims, so they are recorded and attributed, not scored. Most were screened out: both AI providers read the claim on its own and found it a value judgment, rhetoric, hedged, too vague to test, or about something no public evidence could show. The rest were found to be recommendations or too open to test in a full review.
Page 13 of 24: claims 1201 to 1300 of 2353, ordered by the date first said, newest first. To find a claim by its words, use search.
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Newton is of course right that we need to prepare for AGI, whenever it comes.”
Our reading: Society needs to prepare for AGI, whenever it arrives.
No horizon
Qualifier: whenever AGI comes
Source: Hard-forked! Casey Newton’s distorted portrait of Gary Marcus and AI skepticism
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“If there is even a 5% chance I am wrong about diminishing returns, we still ought to be moving immediately to prepare.”
Our reading: Even a 5% chance that LLM returns are not diminishing means society should move immediately to prepare for advanced AI.
No horizon
Qualifier: if there is even a 5% chance he is wrong about diminishing returns
Source: Hard-forked! Casey Newton’s distorted portrait of Gary Marcus and AI skepticism
Assessment: Can’t be checkedSaid
Forecast
“Likewise, an AI with a known tendency to hallucinate things can be used to target weapon (this as may happen soon, given the recent announcement that OpenAI will collaborate with Anduril).”
Our reading: Hallucination-prone AI may soon be used to target weapons, given OpenAI's collaboration with Anduril.
No horizon
Qualifier: given OpenAI's collaboration with Anduril
Source: Hard-forked! Casey Newton’s distorted portrait of Gary Marcus and AI skepticism
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“We need, for example, a global treaty, and those tend to take a decade. Let’s get moving, people.”
Our reading: A global treaty is needed, as one example of preparation for more dangerous AI, and work on it should start now because such treaties tend to take a decade.
No horizon
Qualifier: for example
Source: Hard-forked! Casey Newton’s distorted portrait of Gary Marcus and AI skepticism
Assessment: Can’t be checkedSaid
Conditional forecast
“If AGI dropped tomorrow, given the almost total lack of regulation with real teeth, I seriously doubt we as a society would be ready for it.”
Our reading: If AGI arrived immediately, society would very likely not be ready for it, given the near-absence of regulation with real teeth.
No horizon
Only if: if AGI arrived tomorrow
Source: Hard-forked! Casey Newton’s distorted portrait of Gary Marcus and AI skepticism
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“LLM’s in particular are wildly overrated.”
Our reading: LLMs in particular are wildly overrated.
No horizon
Scope: LLMs in particular, as distinct from AI generally
Source: Hard-forked! Casey Newton’s distorted portrait of Gary Marcus and AI skepticism
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“I also always say we need hybrid systems that include LLMs as one tool among many, never that that LLMs will altogether disappear, even if they become a commodity and valuations fall.”
Our reading: AI needs hybrid systems that include LLMs as one tool among many.
No horizon
Source: Hard-forked! Casey Newton’s distorted portrait of Gary Marcus and AI skepticism
Assessment: Can’t be checkedSaid
Conditional forecast
“But, left unregulated, it probably will.”
Our reading: If AI is left unregulated, it will probably lead to dystopia.
No horizon
Only if: if AI is left unregulated
Source: Two visions of AI’s future
Assessment: Can’t be checkedSaid
Observation
“And it’s not just United; one of the biggest applications of AI seems to be ripping off customers.”
Our reading: It is not just United: one of the biggest applications of AI seems to be ripping off customers.
No horizon
Qualifier: hedged: 'seems to be'
Source: Two visions of AI’s future
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“At the end of the Ezra Klein interview, I called for a massive investment in new approaches to AI. I still think that is the way to go.”
Our reading: He still thinks a massive investment in new approaches to AI, as he called for at the end of the Ezra Klein interview, is the way to go.
No horizon
Source: ChatGPT, at Age Two
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“If you have anything at all private or sensitive in Office documents, you might want to do the same.”
Our reading: Office users who have anything at all private or sensitive in their Office documents might want to switch off the 'optional connected experiences' setting, as Marcus did.
No horizon
Scope: if you have anything at all private or sensitive in Office documents
Source: PSA: Microsoft may be training on your private data without your knowledge
Assessment: Can’t be checkedSaid
Observation
“Every big tech company is desperate for training data, and we can expect more companies following the same sly opt-out playbook, hiding what is going in obscure settings and Terms of Service nobody reads.”
Our reading: Every big tech company is desperate for AI training data.
No horizon
Source: PSA: Microsoft may be training on your private data without your knowledge
Assessment: Can’t be checkedSaid
Rhetoric
“The key point of my recent book Taming Silicon Valley is directly relevant here: none of us stand any chance against big tech unless we stand together.”
Our reading: None of us stand any chance against big tech unless we stand together.
No horizon
Qualifier: unless we stand together
Source: PSA: Microsoft may be training on your private data without your knowledge
Assessment: Can’t be checkedSaid
Rhetoric
“If we don’t, they will walk all over all of us.”
Our reading: If we don't stand together, big tech will walk all over all of us.
No horizon
Qualifier: if we don't stand together
Source: PSA: Microsoft may be training on your private data without your knowledge
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“I would appreciate further, on-the-record clarification from a named member of the company, resolving the appearance of possible contradiction.”
Our reading: Marcus says he would appreciate further on-the-record clarification from a named Microsoft employee, resolving the appearance of possible contradiction between the spokesperson's denial that Microsoft 365 customer data trains large language models and Microsoft's documentation about connected experiences analysing data.
No horizon
Qualifier: on-the-record, from a named member of the company
Assessment: Can’t be checkedSaid
Observation
“Absent a clear statement in the Terms of Service, I personally remain uncertain.”
Our reading: Without a clear statement in Microsoft's Terms of Service, Marcus says he personally remains uncertain whether Microsoft uses customer data from Microsoft 365 to train large language models.
No horizon
Qualifier: absent a clear statement in the Terms of Service
Assessment: Can’t be checkedSaid
Observation
“Third, scaling laws aren’t really laws anymore (if they ever were). The meaning of the word scaling has been greatly devalued.”
Our reading: Scaling laws aren't really laws anymore, if they ever were, and the meaning of the word 'scaling' has been greatly devalued.
No horizon
Qualifier: if they ever were
Assessment: Can’t be checkedSaid
Rhetoric
“Trying to rescue LLMs by making a system that is inherently blind to factuality run longer — rather than better—is like smearing water around wet plates.”
Our reading: Making LLMs, which are inherently blind to factuality, run longer rather than better is a futile way to rescue them.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“We can and must do better. Scaling can only take us so far; the time has come for fresh invention.”
Our reading: The AI field must move beyond scaling and pursue genuinely new ideas.
No horizon
Assessment: Can’t be checkedSaid
Observation
“My guess is o1’s test time scaling works best in closed domains where you can create a lot of reliable synthetic data (such as math problems), and not so well in open-ended domains.”
Our reading: 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.
No horizon
Qualifier: my guess
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“I suggest you keep your hands on your wallets here.”
Our reading: People should be financially wary of the new claims about multiple AI scaling laws.
No horizon
Assessment: Can’t be checkedSaid
Observation
“Scaling can only take us so far; the time has come for fresh invention.”
Our reading: Scaling alone has limited ability to advance AI.
No horizon
Assessment: Can’t be checkedSaid
Observation
“In recent weeks, the third stage has finally arrived.”
Our reading: In recent weeks, as of November 2024, Marcus's 2022 conjecture has reached the stage of being regarded as self-evident, with Microsoft CEO Satya Nadella's remarks as an example. The conjecture is that LLM scaling laws are empirical observations that may not hold and that new approaches would be needed.
No horizon
Scope: in recent weeks
Source: Satya Nadella and the three stages of scientific truth
Assessment: Can’t be checkedSaid
Causal claim
“Almost a half decade lost, when we could have been seriously investing in other approaches and might have made genuine progress towards robust, trustworthy artificial intelligence.”
Our reading: Focusing on LLMs cost AI research almost half a decade. Investing in alternatives might have produced real progress toward robust, trustworthy AI.
No horizon
Qualifier: if other approaches had been seriously funded
Source: Satya Nadella and the three stages of scientific truth
Assessment: Can’t be checkedSaid
Causal claim
“What’s at stake? Science gets delayed.”
Our reading: Resistance to criticism of scaling slowed scientific progress in AI.
No horizon Evidence as of
Limitation: Causal counterfactual claims of this kind cannot be scored from public records.
Source: Satya Nadella and the three stages of scientific truth
Assessment: Can’t be checkedSaid
Forecast
“The point was that we probably needed new architectures.”
Our reading: Marcus restates the point of his 2022 essay: AI probably needed new architectures beyond scaling LLMs.
No horizon
Qualifier: probably
Source: Satya Nadella and the three stages of scientific truth
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Gary Marcus hopes he has taught you by now to never trust the hype.”
Our reading: Readers should never trust AI hype, such as claims that ChatGPT writes poetry indistinguishable from Shakespeare.
No horizon
Source: On hype, and the unbearable banality of ChatGPT’s poetry
Assessment: Can’t be checkedSaid
Observation
“As I wrote yesterday (and first noted back in April), I strongly suspect that, contra Altman, we have in fact reached a point of diminishing returns for pure scaling.”
Our reading: Marcus strongly suspects that pure scaling of AI models has reached a point of diminishing returns, as of November 2024.
No horizon
Assessment: Can’t be checkedSaid
Conditional forecast
“If the rumor is verified, there could be the AI equivalent of a bank run.”
Our reading: If the rumor that a second major AI company is seeing diminishing returns from scaling is verified, the AI industry could see the equivalent of a bank run.
No horizon
Only if: if the rumor that a second major company hit diminishing scaling returns is verified
Assessment: Can’t be checkedSaid
Observation
“Just adding data and compute and training longer worked miracles for a while, but those days may well be over.”
Our reading: The era when simply adding data and compute produced large gains may well be over.
No horizon
Assessment: Can’t be checkedSaid
Forecast
“Reliable, trustworthy AI is surely achievable, but we may need to go back to the drawing board to get there.”
Our reading: Reliable, trustworthy AI is achievable, but getting there may require new approaches beyond current LLMs.
No horizon
Source: CONFIRMED: LLMs have indeed reached a point of diminishing returns
Assessment: Can’t be checkedSaid
Conditional forecast
“When everyone realizes this, the financial bubble may burst quickly; even NVidia might take a hit, when people realize the extent to which its valuation was based on a false premise.”
Our reading: Once people realize LLM economics don't work, the AI financial bubble may burst quickly and even Nvidia might take a hit.
No horizon
Only if: when everyone realizes LLM economics don't work
Source: CONFIRMED: LLMs have indeed reached a point of diminishing returns
Assessment: Can’t be checkedSaid
Conditional forecast
“If LLMs won’t get the US to trustworthy AI, and our adversaries invest in alternative approaches, we could easily be outfoxed.”
Our reading: If LLMs don't lead to trustworthy AI and adversaries invest in alternatives, the US could easily be outfoxed.
No horizon
Only if: LLMs won't lead to trustworthy AI
Source: CONFIRMED: LLMs have indeed reached a point of diminishing returns
Assessment: Can’t be checkedSaid
Forecast
“The economics are likely to be grim.”
Our reading: The economics of LLMs are likely to be grim.
No horizon
Source: CONFIRMED: LLMs have indeed reached a point of diminishing returns
Assessment: Can’t be checkedSaid
Forecast
“But that place may be smaller; it is entirely possible that LLMs on their own will never live up to last year’s wild expectations.”
Our reading: LLMs' role may be smaller than hoped, and LLMs on their own may never live up to the wild expectations of 2023.
No horizon
Qualifier: LLMs on their own
Source: CONFIRMED: LLMs have indeed reached a point of diminishing returns
Assessment: Can’t be checkedSaid
Forecast
“LLMs will not disappear, even if improvements diminish, but the economics will likely never make sense: additional training is expensive, the more scaling, the more costly.”
Our reading: LLMs will not disappear even if improvements diminish, but their economics will likely never make sense because additional training is expensive and more scaling is more costly.
No horizon
Qualifier: even if improvements diminish
Source: CONFIRMED: LLMs have indeed reached a point of diminishing returns
Assessment: Can’t be checkedSaid
Forecast
“The US has been putting all its AI eggs in the LLM basket, and that may well prove to be an epic, massive mistake.”
Our reading: The US has been putting all its AI eggs in the LLM basket, and that may well prove to be an epic, massive mistake.
No horizon
Qualifier: may well
Source: CONFIRMED: LLMs have indeed reached a point of diminishing returns
Assessment: Can’t be checkedSaid
Observation
“Any serious scientist or engineer knows you can’t possibly predict the future with that kind of precision, especially when there are so many unknowns.”
Our reading: Nobody can predict AI's future power, chip, and capital needs with the precision Son claimed.
No horizon
Assessment: Can’t be checkedSaid
Causal claim
“To sweeten it all, they get free press for all these outlandish claims, and that drives up the whole thing.”
Our reading: Those making these outlandish claims get free press for them, and that drives up the whole thing, at least for a while.
No horizon
Qualifier: at least for a while
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Nor should they take anything that he or Son said yesterday seriously.”
Our reading: People should not take Musk's or Son's AI predictions from the Riyadh forum seriously.
No horizon
Assessment: Can’t be checkedSaid
Rhetoric
“It’s a made up round number, with no real meaning; pure hype.”
Our reading: Masayoshi Son's figure that artificial superintelligence will be 10,000 times smarter than humans is a made-up round number with no real meaning; pure hype.
No horizon
Assessment: Can’t be checkedSaid
Causal claim
“That’s why so many investors working with other people’s money love plausible stories with big numbers.”
Our reading: Because investors working with other people's money get paid large fees even when investments do not pan out, so many of them love plausible stories with big numbers.
No horizon
Scope: so many investors working with other people's money
Assessment: Can’t be checkedSaid
Observation
“There is just no way that we can predict all this out to 2040 with any certainty.”
Our reading: Even if one believed scaling was infinite and that useful data would not run out, there is just no way to predict all this, including how efficient future models would be in energy and chips, out to 2040 with any certainty.
No horizon
Qualifier: even if you believed scaling was infinite and useful data would not run out
Assessment: Can’t be checkedSaid
Rhetoric
“As I put it on X last night, quantitative vibes are still just vibes.”
Our reading: Quantitative vibes, meaning gut feelings expressed as numbers, are still just vibes.
No horizon
Assessment: Can’t be checkedSaid
Observation
“In my opinion the audience of investors at the forum was being lied to: they were being assured that everything in the field of AI is under control and well plotted out when in fact science is hard and the field is in its infancy.”
Our reading: In Marcus's opinion, the investor audience at the forum was being lied to: they were being assured that everything in AI is under control and well plotted out, when in fact science is hard and the field is in its infancy.
No horizon
Qualifier: in my opinion
Assessment: Can’t be checkedSaid
Causal claim
“The mere belief in this narrative with respect to AGI could motivate profoundly destabilizing actions, like a nuclear first-strike.”
Our reading: Believing in a clash-of-civilizations race for AGI could motivate profoundly destabilizing actions such as a nuclear first strike.
No horizon
Qualifier: belief in an existential clash-of-civilizations narrative about AGI
Source: “End of an Era”: OpenAI’s AGI Readiness Chief Resigns and Team is Disbanded
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“For now, I’ll just say that we should be very wary of letting science guys with toy models of the world dictate international relations.”
Our reading: We should be very wary of letting AI technologists with simplistic world models dictate international relations.
No horizon
Source: “End of an Era”: OpenAI’s AGI Readiness Chief Resigns and Team is Disbanded
Assessment: Can’t be checkedSaid
Rhetoric
“His departure is yet another sign that OpenAI, as we knew it, is dead.”
Our reading: Brundage's departure is another sign that the original OpenAI no longer exists.
No horizon
Source: “End of an Era”: OpenAI’s AGI Readiness Chief Resigns and Team is Disbanded
Assessment: Can’t be checkedSaid
Causal claim
“However, it’s hard not to think that the timing of his exit might have something to do with the recent reports that OpenAI is trying to restructure as a for-profit public benefit corporation (PBC) as a condition of keeping the $6.6 billion it just raised in its latest funding round.”
Our reading: The timing of Brundage's exit may be connected to OpenAI's attempt to restructure as a for-profit PBC.
No horizon
Source: “End of an Era”: OpenAI’s AGI Readiness Chief Resigns and Team is Disbanded
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“We can and should instead approach the challenge of safely building AGI as one of global, rather than national security.”
Our reading: Safely building AGI should be treated as a matter of global security, not national security.
No horizon
Source: “End of an Era”: OpenAI’s AGI Readiness Chief Resigns and Team is Disbanded
Assessment: Can’t be checkedSaid
Forecast
“Especially since there aren’t many regulations on frontier AI systems and that may not change any time soon.”
Our reading: There are few regulations on frontier AI systems, and this may not change soon.
No horizon
Source: “End of an Era”: OpenAI’s AGI Readiness Chief Resigns and Team is Disbanded
Assessment: Can’t be checkedSaid
Conditional forecast
“Without Microsoft’s unwavering support, they could find themselves in deep trouble.”
Our reading: If OpenAI loses Microsoft's unwavering support, it could end up in deep trouble.
No horizon
Only if: without Microsoft's unwavering support
Assessment: Can’t be checkedSaid
Conditional forecast
“If the answer isn’t bigger LLMs, we may have wasted half a decade.”
Our reading: If bigger LLMs are not the answer, the field may have wasted half a decade.
No horizon
Only if: if bigger LLMs are not the answer
Assessment: Can’t be checkedSaid
Conditional forecast
“When I think of OpenAI, I often think of WeWork: charismatic founder, immense valuation, questionable business plan, and the possibility of similar immense deflation in their valuation, if confidence wavers.”
Our reading: Like WeWork, OpenAI has a questionable business plan and its valuation could deflate immensely if confidence wavers.
No horizon
Only if: if confidence wavers
Assessment: Can’t be checkedSaid
Observation
“OpenAI might be screaming success - but its success is hardly assured.”
Our reading: OpenAI's success is far from assured.
No horizon
Assessment: Can’t be checkedSaid
Conditional forecast
“If things do fall apart, it is not just investors who stand to lose, but society. Immense resources may end up being wasted in vain, because of hype.”
Our reading: If things with OpenAI do fall apart, society as well as investors stands to lose, and immense resources may end up being wasted in vain because of hype.
No horizon
Only if: if things do fall apart
Assessment: Can’t be checkedSaid
Forecast
“Living up to OpenAI’s $150B valuation won’t be easy.”
Our reading: OpenAI will find it hard to live up to its $150B valuation.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“AI ought to be looking itself in the mirror, too, right about now.”
Our reading: The AI industry should take a hard look at its own hype, as biomedicine had to after Theranos.
No horizon
Assessment: Can’t be checkedSaid
Observation
“Nothing that has happened since changes that view.”
Our reading: Marcus says nothing since his February 2023 essay has changed his view that ChatGPT may come to be seen as the greatest publicity stunt in AI history, a glimpse of a future that may take years to realize, with ethical guardrails that will take years to perfect.
No horizon
Scope: since the February 2023 essay
Source: When it comes to security, LLMs are like Swiss cheese — and that’s going to cause huge problems
Assessment: Can’t be checkedSaid
Rhetoric
“It’s a disaster in the making.”
Our reading: The state of LLM chatbot safety is a disaster in the making.
No horizon
Source: When it comes to security, LLMs are like Swiss cheese — and that’s going to cause huge problems
Assessment: Can’t be checkedSaid
Causal claim
“The possibilities are now endless for propaganda, troll farms, and rings of fake websites that degrade trust across the internet.”
Our reading: LLM chatbots open wide possibilities for propaganda, troll farms and fake websites that erode trust online.
No horizon
Qualifier: reaffirmed as "Still true"
Source: When it comes to security, LLMs are like Swiss cheese — and that’s going to cause huge problems
Assessment: Can’t be checkedSaid
Forecast
“In hindsight, ChatGPT may come to be seen as the greatest publicity stunt in AI history, an intoxicating glimpse at a future that may actually take years to realize—kind of like a 2012-vintage driverless car demo, but this time with a foretaste of an ethical guardrail that will take years to perfect.”
Our reading: ChatGPT may come to be seen as AI's greatest publicity stunt, and its ethical guardrails will take years to perfect.
No horizon
Qualifier: reaffirmed from February 2023 essay
Source: When it comes to security, LLMs are like Swiss cheese — and that’s going to cause huge problems
Assessment: Can’t be checkedSaid
Causal claim
“What I argued in 2001, in The Algebraic Mind, still holds: symbol manipulation, in which some knowledge is represented truly abstractly in terms of variables and operations over those variables, much as we see in algebra and traditional computer programming, must be part of the mix.”
Our reading: As Marcus argued in The Algebraic Mind in 2001, symbol manipulation, in which some knowledge is represented truly abstractly in terms of variables and operations over them as in algebra and traditional programming, must be part of the mix.
No horizon Evidence as of
Limitation: The packet cites no evidence about whether current AI systems use symbolic components, or how well systems without them perform.
Source: LLMs don’t do formal reasoning - and that is a HUGE problem
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Everyone actively working with AI should read it”
Our reading: Everyone actively working with AI should read the new Apple researchers' article on LLMs, or at least senior author Mehrdad Farajtabar's X thread summarizing it.
No horizon
Scope: everyone actively working with AI
Source: LLMs don’t do formal reasoning - and that is a HUGE problem
Assessment: Can’t be checkedSaid
Causal claim
“Neurosymbolic AI — combining such machinery with neural networks – is likely a necessary condition for going forward.”
Our reading: Neurosymbolic AI, which combines symbol manipulation with neural networks, is likely a necessary condition for going forward.
No horizon
Qualifier: likely
Source: LLMs don’t do formal reasoning - and that is a HUGE problem
Assessment: Can’t be checkedSaid
Forecast
“(We are, however, unlikely ever to get systematic data on this, since the company isn’t transparent about what it has done or what the results are.)”
Our reading: The public is unlikely ever to get systematic data on robotaxi reasoning failures, because Tesla is not transparent.
No horizon
Qualifier: because Tesla is not transparent
Source: LLMs don’t do formal reasoning - and that is a HUGE problem
Assessment: Can’t be checkedSaid
Forecast
“Elon Musk’s putative robotaxis are likely to suffer from a similar affliction: they may well work safely for the most common situations, but are also likely struggle to reason abstractly enough in some circumstances.”
Our reading: Tesla's planned robotaxis will likely fail to reason abstractly enough in some circumstances, although they may work safely in the most common situations.
No horizon
Qualifier: in some circumstances
Source: LLMs don’t do formal reasoning - and that is a HUGE problem
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Hassabis’s win (shared with DeepMind researcher John Jumper), in contrast, is a slam dunk.”
Our reading: Unlike Hinton's prize, the chemistry Nobel won by Hassabis and shared with John Jumper is a slam dunk, meaning clearly deserved.
No horizon
Assessment: Can’t be checkedSaid
Observation
“Hinton has certainly had a profound influence on machine learning, but it’s still not entirely clear what he specifically won the prize for, or how that has advanced physics.”
Our reading: Hinton has certainly had a profound influence on machine learning, but it is still not entirely clear what he specifically won the physics Nobel Prize for, or how that work has advanced physics.
No horizon
Qualifier: not entirely clear
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“It’s time for a new approach, and Hassabis sees that.”
Our reading: The AI field should move to a new approach beyond pure end-to-end neural networks, as Hassabis recognises.
No horizon
Assessment: Can’t be checkedSaid
Forecast
“His open-mindedness will serve the field well.”
Our reading: Hassabis's open-mindedness will serve the AI field well.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“AlphaFold is a huge contribution to both chemistry and biology. It arguably still hasn’t lived up to the extreme hype that I have sometimes seen, but it’s a fantastic contribution, used widely by biologists, and in my mind one of the two biggest contributions of AI to date, if not the biggest.”
Our reading: AlphaFold is a huge, fantastic contribution to both chemistry and biology and is used widely by biologists. It arguably still has not lived up to the extreme hype Marcus has sometimes seen. In his mind it is one of the two biggest contributions of AI to date, if not the biggest.
No horizon
Qualifier: in my mind
Assessment: Can’t be checkedSaid
Forecast
“People will probably be asking questions about this particular award for a long time.”
Our reading: People will probably keep questioning Hinton's Nobel award for a long time.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Diverting funding from them to more focused efforts around special-purpose AI addressing climate change might make more sense.”
Our reading: It might make more sense to divert funding from chatbots and video synthesis to special-purpose AI addressing climate change.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“One key assumption Schmidt makes is that AI will be eventually worth essentially arbitrary environmental costs. For now, this seems very premature.”
Our reading: For now, Schmidt's assumption that AI will eventually be worth essentially arbitrary environmental costs seems very premature.
No horizon
Qualifier: for now
Assessment: Can’t be checkedSaid
Observation
“All of the major potential virtues are quite speculative.”
Our reading: All the major potential benefits of AI are quite speculative.
No horizon Evidence as of
Limitation: Evidence of research utility does not establish the eventual scale or net societal value of the potential benefits [e1].
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Schmidt needless to say has a large stake in the companies in building in AI, and I think we need to take those potential conflicts of interest seriously.”
Our reading: Schmidt's large stake in AI companies creates potential conflicts of interest that should be taken seriously when weighing his views on AI and climate.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Chatbots and movie synthesis machines (perhaps the most energetically costly systems) are fun, maybe even amazing - but I don’t think we should be risking our planet for them.”
Our reading: We should not risk the planet for chatbots and video synthesis systems.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“It shouldn’t be up to a few billionaires to decide, and profit-driven companies with unelected leaders shouldn’t decide on their own, for all of us. Society has to have a vote here.”
Our reading: Decisions on AI's environmental bet should not be left to billionaires or profit-driven companies; society must have a vote.
No horizon
Assessment: Can’t be checkedSaid
Observation
“The case that AI will do serious harm to the environment if we continue on the current path is actually much stronger.”
Our reading: The case that AI will do serious harm to the environment if we continue on the current path is much stronger than the case for AI's putative benefits.
No horizon
Qualifier: if we continue on the current path
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“For the love of all that is beautiful on this planet, I hope that we think extremely carefully before going literally all in with the Earth on an unproven technology.”
Our reading: Marcus hopes we will think extremely carefully before going literally all in with the Earth on an unproven technology (AI).
No horizon
Qualifier: worded as a hope
Assessment: Can’t be checkedSaid
Observation
“The case for techno-utopia , especially with current technology —which struggles with reasoning in open-ended domains, especially in rare situations with smaller amounts of specific data — is so far very abstract and vague.”
Our reading: Current AI struggles with reasoning in open-ended domains, especially rare data-poor situations, and the case for techno-utopia is so far abstract and vague.
No horizon
Qualifier: especially in rare situations with smaller amounts of specific data
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“We need independent scientists asking about whether the best (rather than simply most conveniently at hand) technologies would be used, and what the odds are of success, and the costs are of any major bets, so we can make a careful decision about what it is worth it.”
Our reading: Independent scientists should assess the technologies, odds of success, and costs of major AI bets before they are made.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Ultimately, we can’t, or at least shouldn’t, place massive bets like these without transparency and accountability.”
Our reading: Massive bets on AI's environmental costs should not be made without transparency and accountability.
No horizon
Assessment: Can’t be checkedSaid
Forecast
“GenAI, which is consuming all the power, probably isn’t going to do a whole lot for addressing climate change, when it can’t even reason reliably about river crossing problems.”
Our reading: GenAI, which is consuming all the power, probably isn't going to do a whole lot for addressing climate change, given that it can't even reason reliably about river crossing problems.
No horizon
Qualifier: probably
Assessment: Can’t be checkedSaid
Observation
“A good IDE (versus none) probably is a much bigger, much less expensive, much less hyped improvement that helps more people more reliably.”
Our reading: Having a good IDE, compared with having none, is probably a much bigger, much less expensive, and much less hyped improvement than generative AI coding tools, one that helps more people more reliably.
No horizon
Scope: a good IDE compared with no IDE
Source: Sorry, GenAI is NOT going to 10x computer programming
Assessment: Can’t be checkedSaid
Causal claim
“10x-ing requires deep conceptual understanding – exactly what GenAI lacks.”
Our reading: Generative AI cannot 10x programming because doing so requires deep conceptual understanding, which GenAI lacks.
No horizon
Source: Sorry, GenAI is NOT going to 10x computer programming
Assessment: Can’t be checkedSaid
Causal claim
“Yet another earlier study (with earlier models) showed evidence of users writing less secure code, which of course could lead to a net loss of productivity long term.”
Our reading: Less secure code written with AI coding assistants could lead to a net loss of productivity in the long term.
No horizon
Qualifier: users writing less secure code
Source: Sorry, GenAI is NOT going to 10x computer programming
Assessment: Can’t be checkedSaid
Causal claim
“Writing lines of code without that understanding can only help so much.”
Our reading: Generating code without conceptual understanding provides only limited productivity help.
No horizon
Qualifier: without conceptual understanding
Source: Sorry, GenAI is NOT going to 10x computer programming
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Use it to type faster, not as a substitute for clear thinking about algorithms + data structures.”
Our reading: Programmers should use generative AI to type faster, not as a substitute for clear thinking about algorithms and data structures.
No horizon
Source: Sorry, GenAI is NOT going to 10x computer programming
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Nobody should able to make that switch for free, at no cost, or people will exploit nonprofit rules infinitely, going forward.”
Our reading: No organization, including OpenAI, should be allowed to switch from nonprofit to for-profit at no cost.
No horizon
Source: How much should OpenAI’s abandoned promises be worth?
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Investors shouldn’t be pouring more money at higher valuations, they should be asking what is going on.”
Our reading: Investors should not put more money into OpenAI at higher valuations and should instead scrutinize what is going on at the company.
No horizon
Source: OpenAI’s slow-motion train wreck
Assessment: Can’t be checkedSaid
Forecast
“This one, from last September, may soon become just as iconic.”
Our reading: The magazine cover of OpenAI's leaders from last September may soon become as iconic as the most famous magazine covers of our time.
No horizon
Scope: the cover from last September
Source: OpenAI’s slow-motion train wreck
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Gary Marcus continue to a dream of day in which AI research doesn’t center almost entirely around LLMs..”
Our reading: AI research should not center almost entirely on LLMs, as it currently does.
No horizon
Assessment: Can’t be checkedSaid
Observation
“(An additional point I should have made is that the new system appears to depend heavily on synthetic data, and that such data may be easier to produce in some domains (such as those in which o1 is most successful, like some aspects of math) than others.)”
Our reading: o1 appears to depend heavily on synthetic data, and such data may be easier to produce in some domains (such as some aspects of math, where o1 is most successful) than in others.
No horizon
Qualifier: appears to depend heavily on synthetic data
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“In an ideal world, both candidates would come clean on their policies around AI, and we should come away with a clear understanding of their understanding of our future. In an ideal world, moderators would demand such clarity.”
Our reading: Presidential candidates should state their AI policies clearly and debate moderators should demand that clarity.
No horizon
Qualifier: in an ideal world
Source: What will the candidates say about AI in tonight’s debate?
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“It would be a really good time to demand better. If we don’t, future generations may regret it.”
Our reading: Citizens should now demand better from candidates on AI policy, or future generations may regret it.
No horizon
Qualifier: if we don't, future generations may regret it
Source: What will the candidates say about AI in tonight’s debate?
Assessment: Can’t be checkedSaid
Forecast
“And how the candidates plan to deal with a likely rise in market manipulation using AI-generated misinformation.”
Our reading: Market manipulation using AI-generated misinformation is likely to rise.
No horizon
Source: What will the candidates say about AI in tonight’s debate?
Assessment: Can’t be checkedSaid
Forecast
“With so much at stake, a total neglect of AI policy would be deeply unfortunate. Our long-term future may actually be shaped more by AI policy than tariffs.”
Our reading: AI policy may shape our long-term future more than tariff policy will.
No horizon
Source: What will the candidates say about AI in tonight’s debate?
Assessment: Can’t be checkedSaid
Forecast
“But what you should really be watching for are the candidate’s comments on AI, which after all, is likely to change the world in coming years, affecting virtually every aspect of society, from employment to education to healthcare to national defense.”
Our reading: AI is likely to change the world in coming years, affecting virtually every aspect of society including employment, education, healthcare and national defense.
No horizon
Source: What will the candidates say about AI in tonight’s debate?