In the Loop: Justine Cassell

Justine Cassell is the new director of the Human-Computer Interaction Institute in the School of Computer Science. A member of the faculty at Northwestern University from 2003 to 2010, she was the founding director of its Center for Technology and Social Behavior, and before that was a tenured professor in the MIT Media Lab.

A graduate of France's Université de Besançon and Dartmouth College, Cassell earned a master's degree in linguistics from the University of Edinburgh and a dual doctorate in linguistics and developmental and cognitive psychology from the University of Chicago.

Credited with developing the Embodied Conversational Agent--a virtual human capable of interacting with people using both verbal and non-verbal behavior--Cassell held the AT&T Research Chair at Northwestern beginning in 2006 and was honored in 2008 with the "Women of Vision" award from the Anita Borg Institute.

In August, she joined SCS as the newest director of the HCII. Cassell spoke to Link Managing Editor Jason Togyer.

How did you get from studying literature and linguistics to analyzing human-computer interaction?

Sometimes I think of it as having been "en-scienced"--kind of like being "entranced." When I began to study literature, I wanted to study people's natural storytelling ability. As a young faculty member at Penn State, I got a grant from the National Science Foundation to buy the very first piece of video capturing software so that I could study how people told stories, and compare those with film they had just watched.

Increasingly, I became interested in the technology not just as a tool for doing my research, but also as a research interest in itself. I received a grant to collaborate with folks at the University of Pennsylvania and build a virtual human and compare different theories of the ways that people tell stories using verbal and non-verbal cues.

At the end of the year I realized I could return to Penn State and my job as a professor of French, linguistics and psychology, or I could explore this new field. That's when I became a human-computer interaction scholar, and I never went back.

Your work on a virtual human resulted in development of the Embodied Conversational Agent. How might a virtual human or ECA complement the research being done at the Pittsburgh Science of Learning Center--particularly in the ongoing development of technologies such as the Cognitive Tutor?

ECAs can and do complement pedagogical agents. The only issue is that they have to be helpful in the experience. The non-verbal behavior has to have some intrinsic benefit.

Sometimes, people stick faces or bodies on things because they think it's going to be "engaging" or "cute." What we know is that while it may hold a student's interest the first time they use the system, it's not going to in the future. To create a really useful virtual human, you have to understand how people communicate, and integrate those aspects that play a role in learning into the pedagogical agent.

How can technology hamper human-to-human interaction, and what strategies can we use to tackle that?

Technology is not a magic bullet, and for any piece of hardware or software that does good for the world or improves communication, there's a piece of hardware or software that hampers communication or can be used for ill.

One of the goals of the Human-Computer Interaction Institute is to understand how to increase the ratio of good technology to bad technology. In some cases, we do that by examining technologies that already exist--when it comes to crowd-sourcing, for instance, the work of Niki Kittur comes to mind, along with the early work done by Sara Kiesler and Bob Kraut to study the social impact of the Internet.

We also do very careful and long-term studies on the relationship and integration of these technologies into the fabric of communities, wherever they've taken place. By doing that kind of research and iteratively re-designing these technologies, the field of human-computer interaction improves the chances that technology will help communication and collaboration, and improve opportunities for learning and the kinds of learning that take place. In this context, Matt Kam's work is a stellar example.

What do you see as near-term future challenges for researchers in human-computer interaction?

How can technology help the elderly have a good quality of life as they age? How do we make sure that all of society--including people with developmental and physical disabilities--gets an equal chance at a good life? How can we make sure that all children get equal access to learning, regardless of their socioeconomic status?

Here, I'm thinking of some of the work that Ken Koedinger, Matt Kam and Vincent Aleven have done with educational software, some of my own work on virtual peers for children with autism, along with the research Jodi Forlizzi and Sara Kiesler have done on using robots in retirement homes, and that Dan Siewiorek has done on mobility technologies through the Quality of Life Technology Center.

How can we use technology to ensure sustainability--and what does it mean to think about technology for a green society? Here, I think of the work Jen Mankoff and Eric Paulos have done.

All of these are examples of the way the HCII works with other parts of SCS to make sure the technologies we develop are used for good, and that those technologies are as usable as they can be and as likely to be used.

Isn't that a broader challenge for students beyond those in HCII?

I'd like to see every student in SCS take at least one HCII class as a way of becoming a better technologist. You can't understand how systems work without understanding how humans interact with each other, and with computers.

One of the extraordinary things about CMU is that Allen Newell and Herb Simon realized from the very beginning that human-computer interaction was a key to all of computer science--that computers need to be embedded in the human world.

What attracted you to the SCS faculty--and to taking on the position of department head at HCII?

What wouldn't attract me? HCII has a stellar faculty in a stellar school of computer science, and the school is arranged such that there are so many people whom I would love to collaborate with in other departments as well. There's a Machine Learning Department, there's a Language Technologies Institute, there's a Robotics Institute--all of the kinds of people I work with are clustered at CMU.
In addition, we get really fabulous, passionate students--they're passionate about building interfaces, but also passionate about the human factor in design. That's one of the things I hope to continue to promote at Carnegie Mellon.
For More Information: 
Jason Togyer | 412-268-8721 | jt3y@cs.cmu.edu