25 years of Intelligent Agent Progress – From the Knowledge Navigator to Siri, Watson, and beyond

For 25 years, since the vision of the Knowledge Navigator was first unveiled in 1987, progress in the field of intelligent agent development has been a utopian goal, with progress occurring sporadically. Within the last year, however, much fanfare has been made about the emergence of several useful agents which provide real utility and provide evidence of great strides being made in the field. The emergence of true “intelligent” agents began with the all important CALO (Cognitive Agent that Learns and Organizes) research at DARPA (Defense Advanced Research Projects Agency) in 2003. This effort was designed to produce artificially intelligent agents that could reason, learn from experience, be told what to do, explain what they were doing, reflect on their experience, and respond robustly to surprise---quite an ambitions task. Today, several advanced Intelligent Agents, while not quite ready to totally automatically ‘learn’ as we do, are providing real value in finding, filtering, and fusing information. Whether searching, filtering, suggesting alternatives, or carrying out instructions, these agents are providing services heretofore relegated to humans. Running on a variety of platforms, including smartphones, Watson, Siri, and Denise, among others, are already solidifying their place as valuable augmenters of human capabilities. Starting with chatbots, such as Sylvie in 1998, we have followed and stayed involved with the development of agents, including some of our own, that can actually facilitate dealing with the information overload problem. This presentation will update the progress, utility and limitations, including several demonstrations, of the latest developments.