Interaction Challenges for Intelligent Assistants
Papers from the AAAI Spring Symposium
Neil Yorke-Smith, Chair
In an increasingly complex world, a new wave of intelligent artificial assistants have the potential to simplify and amplify our everyday personal and professional lives. Whether robotic embodiments or software processes, these intelligent agents will help us manage our time, budgets, knowledge, and workflow as they assist us in our homes, offices, vehicles, and public spaces.
To realize the vision of truly useful assistants, our assistants must be personalized, aware of our preferences, adapting themselves to our ways as well as to new tasks. They must become our partners, able to engage in collaborative problem solving and decision making. Crucially, they must engender our trust over an extended period of time, because their behaviour will materially affect our interests and well-being, and even our own behaviour.
In all these capabilities, an essential aspect of the success of our assistants is their interaction with us and with other humans and agents in natural ways that are no more obtrusive than necessary. This interaction must be uniform and coherent over the various functions of the assistant, and be sensitive to the interaction conditions and modalities, and the user’s time, mood, and cognitive focus.
Developing such intelligent assistants demands collaboration across disciplines. Designing interaction with them challenges us at the level both of fundamental concepts in human-agent communication and of applied research in system building. Hence, from a multidisciplinary perspective, this symposium identified the critical issues raised by interaction with intelligent assistants, the specific challenges faced, and the current state of the art. The ultimate goal was to progress towards the most useful paradigms, methodologies, and implementations for human interaction with intelligent artificial assistants.