The Computational Gauntlet of Human-Like Learning.

In this paper, I pose a major challenge for AI researchers: to developsystems that learn in a human-like manner. I briefly review thehistory of machine learning, noting that early work made close contactwith results from cognitive psychology but that this is no longer thecase. I identify seven characteristics of human behavior that, ifreproduced, would offer better ways to acquire expertise thanstatistical induction over massive training sets. I illustrate thesepoints with two domains - mathematics and driving - where peopleare effective learners and review systems that address them. Inclosing, I suggest ways to encourage more research on human-likelearning.