Scores by kind of claim
Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.
-
Factual claims
Statements about the present or the past, checked against the record: did they hold up?
79 72 85 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
80 72 88 -
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
Insufficient record: 1 promise resolved, 5 needed for a score.
All types combined: 79 (likely 74 to 84, high confidence), from 102 claims: 64 factual claims, 38 predictions. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Gary Marcus's claims
42 claims
“There is very little moat.”
Gary Marcus, Substack author 1 January 2025 25 AI Predictions for 2025, from Marcus on AI (Marcus on AI)
As of end 2024, AI model makers have very little moat, with many similar models and a price war.
Resolved true The record shows it happened or was so.
What would count as true or false
Judged as of the end of 2024. True: evidence from late 2024 shows several developers offering models of similar capability (the top models clustered closely on major leaderboards) and repeated, substantial price cuts across providers consistent with a price war. Mostly true: the top models were similar and prices fell, but one provider held a modest yet noticeable lead or the price cuts were less sharp than 'war' implies. Mixed: either many similar models existed but there was no clear price competition, or there was price competition while one firm held a clear capability lead. Mostly false: one developer held a clear, recognised technical lead, though prices were falling. False: model makers held strong moats, with distinct capability leads and stable or rising prices.
Evidence
- Top AI Stories of 2024! Agents Rise, Prices Fall, Models Shrink, and more... DeepLearning.AI (The Batch) · 25 December 2024
“AI providers waged a price war to attract paying customers. A leading indicator: From March 2023 to November 2024, OpenAI cut the per-token prices of cloud access to its models by nearly 90 percent even as performance improved, input context windows expanded, and the models became capable of processing images as well as text.”
- This Week in AI: AI is rapidly being commoditized | TechCrunch TechCrunch · 28 August 2024
“For one, there’s not much to set the various flagship models apart in terms of capabilities.”
- Technical Performance Stanford Institute for Human-Centered AI (AI Index Report 2026) · 16 August 2026
“This gap narrowed steadily through 2024 as Google, Anthropic, and others released stronger models.”
- Kingy AI Price Index · October 2024 Kingy AI · 19 October 2026
“The median frontier model cost $3.00 per million tokens, down from $5.25 in September. Google and Mistral AI cut their list prices.”
Reasoning
The Batch's 2024 year-end review says that "AI providers waged a price war" and that OpenAI cut per-token prices by nearly 90% between March 2023 and November 2024 (E1). Kingy AI's October 2024 price index records Google and Mistral AI also cutting list prices (E4), so price cuts came from several providers. On capability, TechCrunch reported in August 2024 that there was "not much to set the various flagship models apart" (E2), and the Stanford AI Index says the gap narrowed steadily through 2024 as Google, Anthropic and others released stronger models (E3). Together, independent sources show both close capability among the top models and a price war across providers by late 2024.
Cite this claim
Gary Marcus, “As of end 2024, AI model makers have very little moat, with many similar models and a price war.” (said 1 January 2025). Verdict: resolved true. Reliability API, claim c_30fc612c7546, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_30fc612c7546
Claims without a verdict
Is a verdict here wrong, or do you speak for Gary Marcus? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.