Research

Many activities of a human organization are well-suited for software agents, which can devote significant resources to perform these tasks, thus reducing the burden on humans. Indeed, teams of such software agents could assist all organizations, including disaster response organizations, corporations, the military, universities and research institutions.

Based on the above vision, we have developed a system called the Electric Elves that applies agent technology in service of the day-to-day activities of the Intelligent Systems Division of USC/ISI. Electric Elves is a system of about 15 agents, including nine proxies for nine people, plus two different matchmakers, one flight tracker and one scheduler running continuously for past several months. This paper discusses the tasks performed by the system, the research challenges it faced and its use of AI technology in overcoming those challenges.

A key contribution of our work is understanding the challenges faced in deploying agents to support organizations. In particular, the complexity inherent in human organizations complicates all of the tasks agents must perform. First, since agents must interact with humans, issues of adjustable autonomy become critical. In particular, agents acting as proxies for people must automatically adjust their own autonomy, e.g., avoiding critical errors, possibly by letting people make important decisions while autonomously making the more routine decisions. Second, to accomplish their goals, agents must be provided reliable access to information. Third, people have a wide variety of capabilities, interests, preferences and engage in many different tasks.

To enable teaming among such people for crisis response or other organizational tasks, agents acting as their proxies must represent and reason with such capabilities and interests. We thus require powerful matchmaking capabilities to match both interests and capabilities. Fourth, coordination of all of these different agents, including proxies, is itself a significant research challenge. Finally, the entire agent system must scale-up:

• It must scale-up in the sense of running continually 24 hours a day 7 days a week (24/7) for months at a time
• It must scale-up in the number of agents to support large-scale human organizations.