Could computers design great user interfaces?

Abstract:
In this keynote, I discuss the possibility of using computational methods to design user interfaces automatically. Algorithms play an essential role in the design of many man-made artefacts, so why not in the design of user interfaces? Computational methods are generally useful in well-defined problems in engineering optimization. They are particularly useful when the search space grows too large to be explored manually. I start by observing that many such problems exist in user interface design, too. Consider, for example, the case of designing a menu, one of the most commonly used and most studied user interface type. The number of possible designs for a menu with only 20 items is 2,432,902,008,176,640,000 -- more than there are stars in the observable universe. Interface designers typically approach problems like this with iterative design, heuristics, and rapid prototyping. The traditional user-centered design process is time-consuming, can only cover few alternatives, and success is highly dependent on skills. Disappointingly, however, it offers no guarantee for the outcome — even the most promising outcomes can and will be beaten. Automating user interface design would carry many benefits. Optimization might guarantee better designs — or even the best possible design, as I show in the talk. Designers could focus on truly novel aspects of design, and even novices could design great interfaces. Designers’ work would change. Instead of generating and trying out a few instances at a time, the designer would define optimization goals, assumptions about the user and use, and constraints, and the computer would explore the best designs. But can the computer “understand” the user at the same level as a human designer? I argue that many times it can. I show how predictive models of user interaction can capture very complex human factors affecting user performance and other outcomes. Throughout the talk I show results from the automated design of keyboards, menus, and gestural input. I present interactive design tools that allow even a novice designer to rapidly explore millions of user interface designs — even when the design task is not completely defined. I conclude with a critical discussion of the limits of what computers can do in user interface design.

About the speaker:
Antti Oulasvirta is an Associate Professor at Aalto University where he leads the User Interfaces research group at the Department of Communications and Networking. He was previously a Senior Researcher at the Max Planck Institute for Informatics and the Cluster of Excellence on Multimodal Computing and Interaction at Saarland university. He received his doctorate in Cognitive Science from the University of Helsinki in 2006, after which he was a Fulbright Scholar at the School of Information in University of California - Berkeley in 2007-2008 and a Senior Researcher at Helsinki Institute for Information Technology HIIT in 2008-2011. During his postgraduate studies in 2002-2003, he was an exchange student at UC Berkeley's Neuropsychology Lab and did an internship at T-Labs in Berlin in 2006. Dr. Oulasvirta serves as an associate editor for International Journal of Human-Computer Studies and Personal and Ubiquitous Computing, and he frequently participates in the paper committees of HCI conferences, including the ACM SIGCHI Conference on Human Factors in Computing Systems (CHI). He was awarded the Best Paper Award at CHI in 2011, the Best Paper Nomination at CHI in 2008, 2009, 2013, and 2014, the Best Note Award in 2011 at MobileHCI, and the Most Influential Paper Award in 2013 at MobileHCI.