Held at the 2009 International Joint Conference on Artificial Intelligence
Monday, July 13, 2009 in Pasadena, CA, USA
This workshop engaged people with ongoing and nascent interest in challenges at the intersection of artificial intelligence (AI) and human-computer interaction (HCI). The workshop brought AI researchers together with researchers who are situated more centrally in the HCI community to explore research directions and opportunities for (1) learning and reasoning to enhance human-computer (and human-robot) interaction, and (2) the potential for embedding and involving people more deeply in the operation of learning and reasoning systems.
Research at this intersection is often limited by a separation of concerns. Attempts to automate problem solving and learning can lead AI researchers to bypass prospects for deeper human involvement in the operation of an intelligent system. It is also not uncommon for HCI researchers to attempt to treat AI methods as black boxes, sometimes abandoning promising applications after only initial and shallow explorations of the learning and reasoning necessary to enable those applications. We believe that considering and iterating on both sides of the equation simultaneously opens new opportunities for deep research agendas in both domains.
One focus of the workshop was leveraging interaction to enhance intelligence. Significant prior research on active learning, for example, explores what questions should be asked as part of an efficient learning process. Another approach is to directly expose a system's model, so that a person can use their knowledge of the problem to manipulate system parameters to improve performance. Opportunities for such approaches arise in many domains, including document retrieval, natural language processing, computer vision, and programming by demonstration systems.
Another focus of the workshop was leveraging intelligence to enhance interaction. There is growing interest in both the AI and HCI communities on the role of machine intelligence in human-computer interaction. Efforts include work on recommender systems, mixed-initiative interaction, message triage, new approaches to search and retrieval, and intelligent assistance. Intelligent systems also enable new types of interaction with the physical world, ranging from recognizing everyday human activities using sensor data from a commodity mobile phones to EMG sensing of muscle activity and EEG sensing of brain activity.
We hope to grow the community focused on enhancing both intelligent systems and interaction. This workshop brought together a set of researchers interested in exploring challenges at the intersection of AI and HCI. We welcomed not only attendees already working at this intersection, but also attendees looking to address such questions. We also solicited attendees from both the AI and HCI research communities, as this conversation can be most effective only when including both communities.
Topics of discussion included (but were not limited to):