25 AI Predictions for 2025, from Marcus on AI
We recorded 37 claims from this post. 18 were checked against evidence. 19 can’t be checked. 19 more are held. Read the post at its source.
Claims from this post (37)
In the order they appear in the post. Each quotation is the speaker’s words in this post. A label such as “Supported” or “Can’t be checked” is the claim’s current assessment; its page has the evidence and the reasons.
Assessment: SupportedSaid
Observation
“There is still no robust solution to hallucinations.”
Our reading: There is still no robust solution to LLM hallucinations.
No horizon Evidence as of
Limitation: The main source is a lab paper describing its own models' limits. Independent leaderboard figures for late 2024 could not be read.
Assessment: SupportedSaid
Observation
“There are many such models now; there is a price war. There is very little moat.”
Our reading: There are many GPT-4 level models, a price war, and very little moat among LLM providers.
No horizon Evidence as of
Limitation: 'Moat' is read as a lasting performance lead. Brand, distribution or user base, which are other possible moats, were not measured.
Assessment: UnresolvedSaid
Observation
“Corporate adoption is far more limited than most people expected, and total profits across all companies (except of course hardware companies like NVidia, which profits from chips rather than models) have been modest at best.”
Our reading: Corporate adoption of AI is far more limited than most people expected, and total profits across all companies, except hardware companies like Nvidia that profit from chips rather than models, have been modest at best.
No horizon Evidence as of
Limitation: US broad-AI adoption is only context for corporate generative-AI adoption.
Assessment: Can’t be checkedSaid
Forecast
“AI will presumably solve all of these someday, of course, but current techniques have thus far not been adequate.”
Our reading: AI will presumably eventually solve compositionality, factuality and reasoning, but current techniques have not been adequate.
No horizon
Qualifier: someday
Assessment: UnresolvedSaid
Observation
“AI-generated misinformation became increasingly widespread, as I warned in December 2022, but the effects have – thus far — been less than I feared. I continue to be quite concerned, and think vigilance continues to be required.”
Our reading: AI-generated misinformation became increasingly widespread, but its effects have thus far been less than Marcus feared; he remains quite concerned and thinks vigilance continues to be required.
No horizon Evidence as of
Limitation: Site discovery counts do not establish audience exposure, the proportion of content that is false, or how much growth reflects monitoring coverage [e6, e7].
Assessment: SupportedSaid
Observation
“My prediction that 2024 would be seen as the year of AI disillusionment was overstated but not entirely off the mark, holding true for some folks but not all, and as such perhaps deserve partial credit.”
Our reading: 2024 was a year of AI disillusionment for some but not all; his prediction was overstated.
No horizon Evidence as of
Limitation: The surveys cover business leaders and employees rather than a representative global public.
Assessment: Can’t be checkedSaid
Observation
“My 2023 prediction that OpenAI might someday be seen as the WeWork of AI has not been confirmed, but they are far from out of the woods.”
Our reading: His 2023 prediction that OpenAI might someday be seen as the WeWork of AI has not been confirmed, but OpenAI is far from out of the woods.
No horizon Evidence as of
Limitation: This finding does not establish that OpenAI was financially secure or that the concern was unreasonable.
Assessment: Can’t be checkedSaid
Observation
“Long term security, especially at high valuations, is this far from assured.”
Our reading: OpenAI's long-term security, especially at high valuations, is far from assured.
No horizon
Qualifier: especially at high valuations
Assessment: UnresolvedSaid
Observation
“The company has not returned any profits, and is not expecting to do so until at least 2029, but has a burn rate in the billions, and still lacks a clear technical moat.”
Our reading: OpenAI has no profits, does not expect any until at least 2029, burns billions, and lacks a clear technical moat.
No horizon Evidence as of
Limitation: The financial evidence concerns estimates and expectations, not audited realized results.
Assessment: UnresolvedSaid
Observation
“My predictions about the technical and economic limits of generative AI have largely been on target, but I was flat out wrong about investors.”
Our reading: His predictions about generative AI's technical and economic limits were largely correct, but he was wrong about investor behavior.
No horizon Evidence as of
Limitation: Only the financing component is independently established here.
Assessment: SupportedSaid
Observation
“Immediately after the February announcement of Sora, I noted Sora’s troubles with physics and predicted that they wouldn’t go away. They haven’t.”
Our reading: Sora's troubles with physics have not gone away since its February 2024 announcement.
No horizon Evidence as of
Limitation: The disclosures do not quantify failure rates or establish that physics performance had not improved [e1, e2].
Assessment: UnresolvedSaid
Observation
“My seven December 2022 predictions about GPT-4 largely held true (number seven still pending) for every model that was released in 2024.”
Our reading: His seven December 2022 predictions about GPT-4 largely held true for every model released in 2024, with prediction number seven still pending.
No horizon Evidence as of
Limitation: Selected benchmarks cannot establish a universal claim.
Assessment: Can’t be checkedSaid
Forecast
“I wouldn’t be shocked if none were reliably solved by the end of the year.”
Our reading: Possibly none of the Marcus-Brundage tasks will be reliably solved by end of 2025.
No horizon
Assessment: SupportedSaid
Forecast
“Hallucinations (which should really be called confabulations) will continue to haunt generative AI.”
Our reading: Through 2025 hallucinations will continue to plague generative AI.
Deadline passed Evidence as of
Limitation: 'Haunt' has no threshold. The verdict rests on the plain reading that the problem stayed material and recognized.
Assessment: SupportedSaid
Forecast
“Reasoning flubs will continue to haunt generative AI.”
Our reading: Through 2025 reasoning errors will continue to plague generative AI.
Deadline passed Evidence as of
Limitation: 'Flubs' and 'haunt' set no threshold, so the verdict rests on a plain reading.
Assessment: SupportedSaid
Forecast
“AI “Agents” will be endlessly hyped throughout 2025 but far from reliable, except possibly in very narrow use cases.”
Our reading: AI agents will be heavily hyped in 2025 but far from reliable except possibly in very narrow use cases.
Deadline passed Evidence as of
Limitation: The result partly depends on reading coding agents as a 'narrow use case'. A reader who treats coding as broad could call this mixed.
Assessment: SupportedSaid
Forecast
“OpenAI will continue to preview products months and perhaps years before they are solid and widely available at an accessible price.”
Our reading: By the end of 2025 OpenAI will have continued to preview products months, and perhaps years, before they are solid and widely available at an accessible price.
Deadline passed Evidence as of
Limitation: The result depends on treating premium Pro-only previews as short of ordinary broad consumer access.
Assessment: Can’t be checkedSaid
Forecast
“o3 has been previewed but not released likely will likely be quite expensive.)”
Our reading: OpenAI's o3 will likely be quite expensive when released.
No horizon
Assessment: UnresolvedSaid
Forecast
“Few if any radiologists will be replaced by AI (contra Hinton’s infamous 2016 prediction).”
Our reading: By end of 2025 few if any radiologists will be replaced by AI.
Deadline passed Evidence as of
Limitation: The strongest retrieved workforce evidence concerns the UK.
Assessment: UnresolvedSaid
Forecast
“Truly driverless cars, in which no human is required to attend to traffic, will continue to see usage limited to a modest number of cities, mainly in the West, mainly in good weather. Human drivers will still make up a large part of the economy.”
Our reading: By the end of 2025 truly driverless cars, in which no human is required to attend to traffic, will continue to see usage limited to a modest number of cities, mainly in the West and mainly in good weather, and human drivers will still make up a large part of the economy.
Deadline passed Evidence as of
Limitation: Operator announcements are not a global deployment census.
Assessment: SupportedSaid
Forecast
“Copyright lawsuits over generative AI will continue throughout the year.”
Our reading: Copyright lawsuits over generative AI will continue throughout 2025.
Deadline passed Evidence as of
Limitation: The records establish continuing litigation, not the merits of infringement allegations.
Assessment: UnresolvedSaid
Forecast
“Less than 10% of the work force will be replaced by AI. Probably less than 5%.”
Our reading: By end of 2025 less than 10%, probably less than 5%, of the workforce will be replaced by AI.
Deadline passed Evidence as of
Limitation: Exposure, announced cuts and firm employment effects do not measure realized gross worker replacement [e1, e2, e3].
Assessment: Can’t be checkedSaid
Observation
“Commercial artists and voiceover actors have perhaps been the hardest hit so far.”
Our reading: Commercial artists and voiceover actors have perhaps been hardest hit by AI so far.
No horizon
Assessment: Can’t be checkedSaid
Forecast
“Technical moat will continue to be elusive. Instead, there will be more convergence on broadly similar models, across both US and China; some systems in Europe will catch up to roughly the same place.”
Our reading: With medium confidence, by the end of 2025 a technical moat will continue to be elusive, with more convergence on broadly similar models across the US and China, and some European systems catching up to roughly the same place.
No horizon
Qualifier: filed under medium confidence
Assessment: Can’t be checkedSaid
Forecast
“Few companies (and even fewer consumers) will adopt o3 at wide scale because of concerns about price and robustness relative to that price.”
Our reading: With medium confidence, by the end of 2025 few companies, and even fewer consumers, will adopt o3 at wide scale because of concerns about price and robustness relative to that price.
No horizon
Qualifier: because of concerns about price and robustness relative to that price
Assessment: Can’t be checkedSaid
Forecast
“Companies will continue to experiment with AI, but adoption to production-grade systems scaled out in the real-world will continue to be tentative.”
Our reading: With medium confidence, by the end of 2025 companies will continue to experiment with AI, but adoption of production-grade systems scaled out in the real world will continue to be tentative.
No horizon
Qualifier: filed under medium confidence
Assessment: Can’t be checkedSaid
Forecast
“2025 could well be the year in which valuations for major AI companies start to fall.”
Our reading: Valuations of major AI companies could start to fall in 2025.
No horizon
Qualifier: filed under medium confidence
Assessment: Can’t be checkedSaid
Forecast
“Sora will continue to have trouble with physics.”
Our reading: With medium confidence, by the end of 2025 Sora will continue to have trouble with physics.
No horizon
Qualifier: filed under medium confidence
Assessment: Can’t be checkedSaid
Forecast
“(Google’s Veo 2 seems to be better but I have not been able to experiment with it, and suspect that changes of state and the persistence of objects will still cause problems;”
Our reading: Marcus suspects, without having tested it, that changes of state and the persistence of objects will still cause problems for Google's Veo 2, though it seems better than Sora.
No horizon
Qualifier: has not been able to experiment with it
Assessment: Can’t be checkedSaid
Forecast
“Neurosymbolic AI will become much more prominent.”
Our reading: With medium confidence, by the end of 2025 neurosymbolic AI will become much more prominent.
No horizon
Qualifier: filed under medium confidence
Assessment: Can’t be checkedSaid
Forecast
“We may well see a large-scale cyberattack in which Generative AI plays an important causal role, perhaps in one of the four ways discussed in a short essay of mine that will appear shortly in Politico.”
Our reading: In 2025 there may be a large-scale cyberattack in which generative AI plays an important causal role.
No horizon
Qualifier: filed under low confidence
Assessment: Can’t be checkedSaid
Forecast
“Instead we may see models like o1 that are quite good at many tasks for which high-quality synthetic data can be created, but in other domains only incrementally better than GPT-4.”
Our reading: In 2025 we may see o1-like models good at synthetic-data-friendly tasks but only incrementally better than GPT-4 elsewhere.
No horizon
Qualifier: tasks where high-quality synthetic data can be created
Assessment: Can’t be checkedSaid
Forecast
“There could continue to be no “GPT-5 level” model (meaning a huge, across the board quantum leap forward as judged by community consensus) throughout 2025.”
Our reading: There could be no GPT-5 level model, meaning a huge across-the-board leap by community consensus, throughout 2025.
No horizon
Qualifier: filed under low confidence
Assessment: SupportedSaid
Forecast
“AI “Agents” will be endlessly hyped throughout 2025 but far from reliable, except possibly in very narrow use cases.”
Our reading: Throughout 2025, AI agents will be endlessly hyped but far from reliable, except possibly in very narrow use cases.
Deadline passed Evidence as of
Limitation: The promotion shown comes from one company, OpenAI. It does not measure how much agents were promoted across the industry.
Assessment: Can’t be checkedSaid
Forecast
“Technical moat will continue to be elusive. Instead, there will be more convergence on broadly similar models, across both US and China; some systems in Europe will catch up to roughly the same place.”
Our reading: Filed under 'Medium confidence': through the end of 2025, a technical moat will continue to be elusive; instead there will be more convergence on broadly similar models across both the US and China, and some systems in Europe will catch up to roughly the same place.
Deadline passed Evidence as of
Limitation: This record does not show what happened in 2025. No evidence about the gap between leading models is part of this record.
Assessment: Can’t be checkedSaid
Forecast
“Few companies (and even fewer consumers) will adopt o3 at wide scale because of concerns about price and robustness relative to that price.”
Our reading: Filed under 'Medium confidence': by the end of 2025, few companies (and even fewer consumers) will adopt OpenAI's o3 at wide scale, because of concerns about price and robustness relative to that price.
Deadline passed Evidence as of
Limitation: This record does not show how widely o3 was adopted in 2025. No evidence on its adoption is part of this record.
Assessment: Can’t be checkedSaid
Forecast
“Sora will continue to have trouble with physics.”
Our reading: Filed under 'Medium confidence': at the end of 2025, OpenAI's Sora will still have trouble with physics.
Deadline passed Evidence as of
Limitation: The only evidence shown is OpenAI's own description, not an independent test.
Held from this post (19)
These records did not pass every check, so nothing about them is published. Held is not a finding.
- The two reviews read the claim differently
- The criteria are ambiguous. Our reading, not assessed: His predictions since 2022 that pure scaling of LLMs would eventually start to run out appear to have been confirmed in November and December 2024, based on comments from many industry insiders.
- The two reviews reached different assessments. Our reading, not assessed: In 2025 humanoid robotics will be hyped but no one will release a robot remotely as capable as Rosie; situational awareness and flexibility will remain poor.
- The two reviews reached different assessments. Our reading, not assessed: Marcus's 2024 AI predictions have largely held, with some caveats.
- The two reviews reached different assessments. Our reading, not assessed: GPT-5 was not released in 2024 despite repeated predictions.
- The criteria are ambiguous. Our reading, not assessed: Weaknesses in compositionality, factuality and reasoning persist in current AI.
- The two reviews reached different assessments. Our reading, not assessed: AGI will not be achieved in 2025.
- The two reviews reached different assessments. Our reading, not assessed: In 2025 the US will continue to have very little regulation protecting consumers from generative AI risks.
- The two reviews read the claim differently
- The two reviews reached different assessments. Our reading, not assessed: In 2025 much of the world will increasingly look to Europe for AI regulatory guidance.
- The two reviews reached different assessments. Our reading, not assessed: LLMs did not markedly improve commercial assistants like Amazon Alexa during 2024.
- The record does not meet the current rules. Our reading, not assessed: Most AI model companies have lost money, and at least three have been quasi-acquired with little profit for investors.
- The two reviews read the claim differently
- The two reviews reached different assessments. Our reading, not assessed: In 2025 AI Safety Institutes will offer guidance but have little legal authority to stop dangerous models.
- The record does not meet the current rules. Our reading, not assessed: Through 2025, profits from AI models will remain modest or nonexistent while chipmakers continue to do well.
- The two reviews reached different assessments. Our reading, not assessed: Available AI models remain essentially GPT-4 level with incremental changes, and OpenAI has nothing it deems worthy of the GPT-5 label.
- The two reviews reached different assessments. Our reading, not assessed: No single AI system will solve more than 4 of the Marcus-Brundage AI 2027 tasks by end of 2025.
- The two reviews read the claim differently
- The two reviews reached different assessments. Our reading, not assessed: In 2025 AI power consumption will rise and remain a major problem, with few generative AI companies transparent about usage.