If no company releases a true quantum leap addressing reliability, hallucination, data leakage and security by the end of 2024, the generative AI bubble may start to pop by around March 2025.
Unresolvable The public record could not settle it.
Prediction: Statements about events outside the person's control, checked once they fell due: did they come true?
The second judge would not resolve this claim; the verdict shown is the first judge's.
What would count as true or false
This is a conditional prediction with strong hedging ('may'). First, check the condition: whether, by 2024-12-31, any company released a system widely recognised as a true leap that substantially solved reliability, hallucination, data leakage and security. If such a leap was released, the condition failed and the claim cannot be resolved. If no such leap was released, judge the outcome by about 2025-03-31: did the generative-AI bubble 'start to pop'? Count as evidence a sustained, significant fall in generative-AI valuations (for example a drop of 20% or more, lasting weeks, in leading AI stocks or AI company valuations), a clear fall in generative-AI venture funding or in big-tech AI capital spending plans, or widespread reports of AI companies failing or being sold at distressed prices. True: the condition was met and clear signs of a bubble deflating, as above, appeared by around March 2025. Mostly true: the condition was met and signs of deflation appeared, but they were short-lived or limited to part of the sector. Mixed: the condition was met and the signals conflicted, for example a sharp but brief market drop while funding and capital spending kept rising. Mostly false: the condition was met, but by March 2025 there were only isolated warning signs, while valuations, funding and spending kept growing overall. False: the condition was met and by around March 2025 generative-AI valuations, funding and investment were stable or rising, with no meaningful deflation. Because of the hedge, an absence of popping weakens the claim but still counts against it as described.
Set before the research began.
Evidence (4)
-
E1 State of AI Q1’25 Report - CB Insights Research CB Insights · 12 September 2025
“AI funding grew 51% to $66.6B across 1,134 deals in Q1’25.”
-
E2 State of AI Q1’25 Report - CB Insights Research CB Insights · 12 September 2025
“AI funding grows 51% QoQ to hit $66.6B — a new quarterly record — as industry incumbents secure mega-rounds.”
-
E3 DeepSeek crashes the AI Party: Story Break, Change or Shift? Aswath Damodaran (blog) · 31 January 2025
“Nvidia Broadcom, Constellation and Vistra have had terrible weeks, losing more than 10% in the last week, but just for perspective, also note that Constellation and Vistra are still up strongly for the year.”
-
E4 DeepSeek crashes the AI Party: Story Break, Change or Shift? Aswath Damodaran (blog) · 31 January 2025
“Meta and Microsoft were unaffected, and so was Palantir, Clearly, the DeepSeek story is playing out differently for different companies in the AI space, but its overall market impact has been substantial, and for the most part, negative.”
Reasoning
None of the evidence addresses whether a 2024 release substantially solved reliability, hallucination, data leakage and security, so the claim's condition cannot be confirmed. The outcome evidence shows record Q1 2025 AI funding [E1][E2] and an uneven late-January 2025 sell-off [E3][E4], but it does not cover valuations through March 2025 or AI capital spending.
Cite this claim
Gary Marcus, “If no company releases a true quantum leap addressing reliability, hallucination, data leakage and security by the end of 2024, the generative AI bubble may start to pop by around March 2025.” (said 31 March 2024). Verdict: unresolvable. Reliability API, claim c_bfa02b5566fd, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_bfa02b5566fd
This claim is in dataset release v1.1.2, with the same verdict.
Reply to this claim
More prediction claims by Gary Marcus
-
Resolved true
“Profits from AI models will continue to be modest or nonexistent”
Profits from AI models will remain modest or nonexistent in 2025.
-
Mostly true
“GPT-5 may be more satisfying but is likely even more expensive to train and operate.”
GPT-5 is likely to be even more expensive to train and operate than GPT-4.
-
Mostly true
“Some of its advice will be head-scratchingly bad.”
-
Mostly true
“GPT-5 will surely be better, a lot better than GPT-4.”
-
Resolved true
“GPT-5 really is about to drop, no foolin’.”
Verdicts are made automatically; how verdicts are made.