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
“my bet: hybrid models will pull ahead next year.”
Gary Marcus, author of the post 14 December 2021 Post by @GaryMarcus on X, 2021-12-14 (X)
Gary Marcus bet that hybrid (neurosymbolic) AI models would pull ahead in 2022.
Resolved false The record shows it did not happen or was not so.
The second judge would not resolve this claim; the verdict shown is the first judge's.
What would count as true or false
True: during 2022, hybrid (neurosymbolic) AI systems clearly outperformed pure deep-learning systems on the main AI benchmarks and among the most prominent and influential AI systems released that year, and contemporary expert reviews of 2022 recognised hybrid approaches as having moved ahead. Mostly true: hybrids led in most major areas but not all. Mixed: hybrids led clearly in some notable domains, while pure deep learning led in others of similar importance. Mostly false: hybrids gained ground or led only in narrow or niche tasks, while pure deep-learning systems led the most prominent benchmarks and releases. False: pure or mostly deep-learning systems dominated state-of-the-art results and the major releases of 2022, and contemporary reviews did not describe hybrid models as having pulled ahead. Results after 2022-12-31 do not count.
Evidence
- Biggest AI Headlines in 2022 – At Stanford and Beyond Stanford HAI · 14 December 2022
“If we distill the biggest advances in AI this year into one headline, it would be generative AI. This year companies launched major models like ChatGPT, DALL-E 2, and others, highlighting impressive capabilities and sparking debate about potential harms.”
- 2023 AI Index: A Year of Technical Achievement, Newfound Public Scrutiny Stanford HAI · 3 April 2023
“Large language models, which have driven much recent AI progress, are getting bigger and more expensive.”
- Technical Performance Stanford HAI · 1 January 2023
“Language models continued to improve their generative capabilities, but new research suggests that they still struggle with complex planning tasks.”
- 2022 in artificial intelligence - Wikipedia Wikipedia · 9 February 2025
“April 12 – DeepMind introduces Chinchilla, a 70-billion-parameter language model that outperforms larger models, including Gopher, on a range of benchmarks.”
Reasoning
Stanford HAI's review of 2022 (E1) names generative AI as the year's biggest advance, citing ChatGPT and DALL-E 2. It says nothing about hybrid or neurosymbolic systems pulling ahead. The 2023 AI Index (E2, E3) credits large language models with driving recent AI progress. It notes only that they still struggle with complex planning. E4 records Chinchilla, a 70-billion-parameter language model, outperforming larger models on a range of benchmarks in 2022. In these independent sources, generative AI and large language models are named as the main advances of 2022, and none describes hybrid models as having moved ahead.
Cite this claim
Gary Marcus, “Gary Marcus bet that hybrid (neurosymbolic) AI models would pull ahead in 2022.” (said 14 December 2021). Verdict: resolved false. Reliability API, claim c_cfcb4af12e3d, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_cfcb4af12e3d
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.