Usable Intelligent Interactive Systems

"The AI and HCI communities have often been characterized as having opposing views of how humans and computers should interact"
observes Winograd in Shifting Viewpoints. It is time to narrow this gap. What was once considered the forefront of artificial intelligence (AI) research can now be found in commercial products. While some have failed, others, such as face detection in digital cameras or product recommendation systems, have become so mainstream they are no longer thought of as artificial intelligence.

This special interest group at CHI 2009 in Boston will provide a forum to examine the apparent gap between HCI and AI communities, to explore how intelligent technologies can enable novel interaction with computation, and to investigate the challenges associated with understanding human abilities, limitations, and preferences in order to drive the design of intelligent interactive systems.

Goal

The intent of this special interest group is to foster a discussion into how HCI and AI collaboration can lead to new developments that integrate the strengths of humans and AI systems. In particular, we are interested in the following questions:

• How can AI increase usability in systems? What general principles of interface design can AI help to satisfy? For instance, all interfaces should accomplish interactions in ways that make efficient use of user input, and use methods of input and output that are natural for the human user. How can intelligent systems expand this space? How can they enable novel ways of interacting with computing systems?
• What are the usability challenges common to interactive intelligent systems, and how can these be mitigated? How do we design robust interactions for systems that include components that do the "right thing" only some of the time? How do we make proactive systems appear predictable? How can we help users form helpful mental models of systems whose behavior is governed by complex statistical inference?
• What are the effective methods for ensuring and improving the usability of intelligent systems? Given that designers are currently tackling problems in the HCI-for-AI design space, how do the existing usability evaluation methods hold up? Which ones work best for which kinds of applications or algorithms? Are there systematic gaps in evaluation methods? For example, what evaluation methods are appropriate for applications with user intent recognition and/or machine learning algorithms? What are the gaps in design theory (computational theory, evaluation theory, interaction design approaches) that would benefit from systematic research? Do AI systems need new design methods or prototyping tools? How do we identify and fill these gaps?

The overall goal of this special interest group is to initiate a community of researchers and practitioners who are interested in this topic and who want to exchange information and experience and form collaborations on specific projects.

The most visible result will be a website for this community, which will be initialized before the SIG meeting and extended after the meeting with the help of participants in the meeting.

The initial plan is for this website to include:

• Links to relevant literature, conferences, and other events;
• Pages for individuals and groups on which they can describe their own experience and interests;
• A discussion forum in which projects and issues of interest to the community can be discussed.

Audience

The audience of this SIG consists of practitioners and researchers in areas that overlap with the intersection human-computer interaction and artificial intelligence, such as: end user programming, knowledge capture, user modeling and adaptivity, robotics, assistive and agent technologies, multimodal systems, cognitive science, and fields that address complex socio-technical systems.

More Information

See the official SIG abstract and the CHI schedule.