In Our Own Image: Building an Artificial Person

Neural networks consultant Caudill (coauthor of Naturally Intelligent Systems) here surveys artificial life research, then reaches beyond it to grab for the hands of the androids she believes might be produced within 20 years. Sketching the likely design of these intelligent machines, she uses broad, heuristic (her favorite word) strokes, primarily dealing with software systems as she examines the technological gaps between current theory and the achievements necessary to make androids a practical possibility. These gaps have virtually disappeared in such areas as vision, speech recognition and mobility, and Caudill argues convincingly that they are closing in fast in all other areas, including the vital one of true intelligence. Although she sometimes goes far afield in positing largely stochastic ethical issues (Should it be legal to have a sexual relationship with an android? At what point do industrial androids become slaves?), she presents an intriguing vision of a future in which human and mechanical evolution will be intertwined. One can almost feel the androids of Isaac Asimov's I, Robot reading over Caudill's shoulder. (Oct.)