LAWS@KL.SRI.COM (Ken Laws) (02/01/88)
I didn't attend the lecture, but have read the Dreyfuses' paper in the new Daedalus issue previously mentioned in AIList. The paper says little about connectionism specifically (except that it may now be getting a deserved chance to fail just as symbolic AI has done), but the authors are favorable to holistic approaches in general and seem less negative about an AI field that includes this emphasis. Their paper and one by Papert were mainly concerned with the history of AI and how the symbolic paradigm gained supremacy. Almost every paper in the Daedalus issue comments on (or indeed focusses on) connectionism and neural models, mainly at a philosophical level of discussion. Sherry Turkle, for instance, discusses connectionism and psychoanalysis as resonant fields that need to initiate a dialog about object-oriented mental models. It's a very interesting collection of essays. -- Ken -------