How do we build rational machines and what does it mean to be rational? This talk will look at the evolution of intelligent software agents, including the early work at SRI International on Shakey the Robot and the beginnings of the modelling of the mental state of software agents in terms of their beliefs, goals and intentions.
What is the importance of these key mental attitudes and how do they affect the way an intelligent agent perceives the world and acts to accomplish its goals? How can an agent recognise the intentions and goals of others, including agents (or people) that are not cooperative? Why do some propose that decision theory won’t work in practice, and why do intelligent machines have to throw away potential solutions if they are going to survive in the real world?
The talk will also look at some of the early applications of these intelligent agents, including managing malfunctions on NASA’s space shuttle and air combat missions, to later business applications and their potential role in the management of people with chronic disease, such as diabetes. Finally, we will consider where these agents will go in the future, and how emotional states may be key to their survival.