50 years into AI’s US history and 25 years into AAAI’s history, we’ve come a long way. There are a wide variety of deployed applications based on AI or incorporating AI ideas, especially applications involving machine learning, data mining, vision, natural language processing, planning and robotics. A large fraction of the most exciting opportunities for research lie on the interdisciplinary boundaries of AI with CS (systems, graphics, theory, etc.), biology, linguistics, engineering and science. Vastly increased computing power has made it possible to deal with realistically large though specialized tasks.
However we are still are far short of truly intelligent systems in the sense that people are intelligent—able to display “common sense”, deal robustly with surprises, learn from anything that can be expressed in natural language, understand natural scenes and situations, etc. At the same time AI has tended to splinter into specialized areas that have their own conferences and journals, and that no longer have the goal of understanding or building truly intelligent systems.
My own sense is that the AI research program needs to be rethought in order to have a realistic hope of building truly intelligent systems, whether these are autonomously intelligent or “cognitive prostheses” for human-centered systems. Early AI focused on the aspects of human thought that were not shared with other creatures—e.g. reasoning, planning, symbolic learning—and minimized aspects of intelligence that are shared with other creatures—e.g. vision, learning, adaptation, memory, navigation, manipulation of the physical world. A truly intelligent system will require an architecture that layers specifically human-like abilities on top of abilities shared with other creatures. Some recent programs at DARPA and NSF are setting ambitious goals that will require integrated generally intelligent systems, a very promising trend. The best news about the neglect of integrated intelligent systems is that researchers going into this area are likely to encounter a good deal of “low hanging fruit”.