I can think of nothing more exciting than working on integrated human-level intelligent systems and after 25 years it is still captivating and challenging. One challenge is that it requires interdisplinarity in the small (across subfields of AI) and large scale (with other disciplines outside of AI). A second challenge is that teams are needed to attack the large scale problems that can challenge integrated AI systems (as evident in many of the recent DARPA programs). This type of research isn’t for the faint at heart or those who enjoy solitary work. A third challenge is to communicate research results—if the work is truly interdisciplinary, which field or subfield should it be published in? Moreover, how (and where) can I talk about what is learned about integration, which itself is not native to any specific field. AI has done a marvelous job, leveraging specialization with the inexhaustible growth of new subfields, new applications, and new conferences; but we also need to fiercely support integration. What better way to do this than by making integrated cognitive systems a major emphasis of the AAAI conference?
As teachers, we must challenge students to work on problems where integration is central and not an after thought: problems that require large bodies of different types of knowledge, problems that involve interaction with dynamic environments, problems that change over time and problems where learning is central (and sometimes problems where determining the appropriate metrics is part of the research). But given the dynamics in the field of AI, a Ph.D. student must forge an association to an identifiable subfield of AI—some community in which to publish and build a reputation—and as of today that is not “human-level intelligence”, “integrated cognitive systems”, or even my favorite, “cognitive architecture”. So even more than finding a home for publishing, we must grow a community of researchers, teachers, and students where the integration is the core and not the periphery.