Our weekly Seminar Series welcomes Matt Ratto, an associate professor in the Faculty of Information at the University of Toronto and the Bell University Labs Chair in Human-Computer Interaction. Ratto’s research explores intersections between digital technologies and the human life world, with a particular focus on new developments that trouble the divide between online and offline modes of production. His work crosses boundaries between the digital and physical and the divide between humanities and engineering disciplines. In 2007, he coined the term “critical making” to describe work that combines humanities insights and engineering practices, and has published extensively on this concept. He is currently developing cost-effective software and hardware for scanning, design, and 3D printing prostheses for use in the developing world, in partnership with Hope and Healing International, and rehabilitation hospitals in Canada, Uganda, and Tanzania.
Talk title
“> and < human? Behavioural intervention and counselling bots”
Abstract
An interdisciplinary team of computer scientists, engineers, clinicians, and researchers from CAMH and the University of Toronto are currently developing an AI chatbot system to engage in conversations with habitual smokers and assist in increasing their motivation to quit. This system is being designed and programmed using techniques from Motivational Interviewing (MI), a client-centred style of psychotherapy that focuses on encouraging clients to voice their own motivations for behavioural change. Part of this work includes open-ended social research looking at general questions regarding the use of bots for behavioural intervention. Of particular interest from both a technical and a social perspective is the quality and goal of the human-bot interactions, the performance and development of trust, and, ultimately, positive counselling outcomes.
In this talk I will summarize the project as it currently stands, describing the development of a previous “motivational” bot and how it differs from the development work that is currently ongoing. Briefly, the primary difference is in the role played by natural language processing (NLP) systems in engaging in dialogue with users. While the initial bot prototypes developed in this project used NLP (specifically a Convolutional Neural Network architecture from spaCy) to classify user expressions and provide boilerplate responses, new versions of the bot leverage the GPT2 and 3 NLP system to classify as well as generate more complex bot responses. As such, the new bot is intended to respond to users in a more “human-like” manner, generating appropriate reflections based on the bot-user dialogue as well as positive affirmations to support behavioural change.
The capacity of the bot to generate human-like responses has raised a productive contradiction within this work. One assumed benefit of a counselling bot rather than a human counsellor is the sense that because the bot lacks human positionality it will be less judgmental. However, it is also typically assumed that the more a bot can simulate human-like characteristics, the more a user will trust the bot and be prepared to engage with it. We have begun exploring this issue, using initial work on trust as a stalking horse for the deeper issue; namely, should counselling bots interact with humans as if they are also human, or is an alternative non-human subject position potentially more productive?
Recommended readings
Almusharraf, F., Rose, J., & Selby, P. (2020). Engaging Unmotivated Smokers to Move Toward Quitting: Design of Motivational Interviewing-Based Chatbot Through Iterative Interactions. Journal of Medical Internet Research, 22(11), e20251. https://doi.org/10.2196/20251
Braidotti, R. (2013). The Posthuman. First edition. Polity.
Tekin, Ş. (2020). Is Big Data the New Stethoscope? Perils of Digital Phenotyping to Address Mental Illness. Philosophy & Technology. https://doi.org/10.1007/s13347-020-00395-7https://doi.org/10.2196/20251
About Matt Ratto
Matt Ratto is an associate professor in the Faculty of Information at the University of Toronto, and the Bell University Labs Chair in Human-Computer Interaction. His research focuses on how theories and perspectives from technoscience research can usefully extend and contextualize design and engineering practice, particularly related to emerging digital technologies. He is also Chief Science Officer of Nia Technologies, a non-profit organization that provides 3D printing resources for rehabilitation clinics in the developing world and Chief Technology Officer for Readout Health.
About the SRI Seminar Series
The SRI Seminar Series brings together the Schwartz Reisman community and beyond for a robust exchange of ideas that advance scholarship at the intersection of technology and society. Seminars are led by a leading or emerging scholar and feature extensive discussion.
Each week, a featured speaker will present for 45 minutes, followed by 45 minutes of discussion. Registered attendees will be emailed a Zoom link approximately one hour before the event begins. The event will be recorded and posted online.
Matt Ratto