Gary Marcus on X: "can't believe people assume that success on highly verifiable problems in math (where we don't even know how many tests were performed and how many might have failed) iautomatically generalize to everything else when there is not a shred of evidence that they do. 🤷♂️" / X
Gary Marcus(@GaryMarcus)151 字 (约 1 分钟)
72
Gary Marcus questions whether AI's success on highly verifiable mathematical problems can automatically generalize to other domains, pointing out lack of evidence for such generalization, while Soham Mehta counters that OpenAI has proven a long-standing conjecture in discrete geometry, challenging the view that AI is merely a stochastic parrot.
入选理由：AI在数学验证问题上的成功缺乏充分测试数据支撑其泛化能力
FeaturedTweet#AI#Mathematical Verification#Machine Learning#OpenAI英文