Autocorrect to Alexa - Understanding the human experience

Originally published on the University of Canterbury website.

So much of what we do today – from texting to choosing music – is underpinned by a new generation of intelligent computer interfacing where our intentions are inferred and output altered in ways we may barely notice. Sophisticated artificial intelligence (AI) is fairly pervasive now, spanning everything from autocorrect and ‘Snap to Grid’ functions through to Alexa or Siri.

Professor Andrew Cockburn, of UC’s Computer Science and Software Engineering department, is leading a research project aimed at understanding the human experience of these intelligent interfaces. Joining him on the project, which is being funded with a three year grant from the 2019 Marsden Fund, are fellow UC researcher Associate Professor Zhe Chen from the School of Psychology, Speech and Hearing, and Professor C.A. Gutwin from the University of Saskatchewan, Canada.

“A wide spectrum of systems today try to guess or infer your intentions as a user and modify outputs based on that,” says Professor Cockburn, who is also the director of UC’s Human Computer Interaction (HCI) Lab. “The user experience is often variable. Autocorrect is the perfect example. If you type ‘Tumaru’ when you meant to type ‘Timaru’ and it makes the correct change then that’s good, but if you type ‘Timaru’ and it gets changed to ‘Tomatillos’ it can be terribly frustrating!”

This project will study the quality of people’s experiences by using lab studies and crowdsourcing sites and explore the extent to which people are willing to put up with irritating outcomes if the overall experience is still a positive one. “How much loss are people willing to trade-off for how much gain? Our theory is that if there’s an element of loss, that’s going to have an over-weighted influence.”

By developing new understandings and predictive models of user experiences, the researchers hope to help designers avoid developing interfaces that will be disliked by users. Professor Cockburn says the Marsden Fund grant gives an opportunity to further build the HCI Lab and UC’s expertise in this area, along with exciting collaborative work.