Modeling Social Action for AI Agents

AI is a science, not merely technology, engineering. It cannot find an identity (ubi consistam) in a technology, or set of technologies, and we know that such an identification is quite dangerous. AI is the science of possible forms of intelligence, both individual and collective. To rephrase Doyle's claim, AI is the discipline aimed at understanding intelligent beings by constructing intelligent systems. Since intelligence is mainly a social phenomenon and is due to the necessity of social life, we have to construct socially intelligent systems to understand it, and we have to build social entities to have intelligent systems. If we want that the computer is not "just a glorified pencil" [Popper, BBC interview), that it is not a simple tool but a collaborator [Grosz, 1995], an assistant, we need to model social intelligence in the computer. If we want to embed intelligent functions in both the virtual and physical environment (ubiquitous computing) in order to support human action, these distributed intelligences must be social to understand and help the users, and to coordinate, compete and collaborate with each other. In fact Social Intelligence is one of the ways AI responded to and went out of its crisis. It is one of the way it is "back to the future", trying to recover all the original challenges of the discipline, its strong scientific identity, its cultural role and influence, that in the '60s and 70s gave rise to the Cognitive Science, and now wil l strongly impact on the social sciences. This stream is part of the new AI of the '90s where systems and models are conceived for reasoning and acting in open unpredictable worlds, with limited and uncertain knowledge, in real time, with bounded (both cognitive and material) resources, with hybrid architectures, interfering -either cooperatively or competitivelywith other systems. The new password is interaction [Bobrow, 1991): interaction with an evolving environment; among several, distributed and heterogeneous artificial systems in a network; with human users; among humans through computers. Important work has been done in AI (in several domains from D A I to HCI , from Agents to logic for action, knowledge, and speech acts) for modeling social intelligence and behavior. In my talk I wi l l just attempt a principled systematization. On the one side, 1 wil l illustrate what I believe to be the basic ontological categories for social action, structure, and mind; letting, first, sociality (social action, social structure) emerge bottom-up from the action and intelligence of individual agents in a common world, and, second, examine some aspects of the way-down: how emergent collective phenomena shape the individual mind. In this paper I wil l focus on the bottom-up perspective. On the other side, I wi l l propose some critical reflections on current approaches and future directions. Doing this I wi l l in particular stress five points.