Full Title: Me, myself and A.I.: How interacting with Artificial Intelligence affects employees’ expectations of, and attitudes towards, the role of A.I.
Speaker:
Meena Andiappan, Ph.D. (she/her)
Assistant Professor of Management & Organization
Institute for Health Policy, Management, and Evaluation
Faculty Affiliate, Schwartz Reisman Institute for Technology and Society
Faculty Affiliate, Joint Center for Bioethics
University of Toronto
Abstract
AI describes a new class of technology, one that is capable of interacting with humans and the environment alike and that strives to mimic human capabilities (Rahwan et al, 2019). Although people have a range of fears and expectations about the role of AI (Artificial Intelligence) in organizations, until very recently, the bulk of these attitudes are based on impressions of AI technology (e.g., gleaned from popular and news media) rather than first-hand knowledge of AI outputs. Given the growing role that experts expect AI to play in both our personal lives and our professional lives (Danaher, 2017; Huang & Rust, 2018; Kaplan, 2015; Kellogg, Valentine & Christin, 2019) and the rapid rate of technological adaption by workplaces (West, 2018), we explore how one-on-one interaction with AI affects employees’ perceptions of AI’s abilities and their willingness to work with AI in the future. We conduct two online studies using human-AI (GPT-3) interactions to test employees’ general and task-related attitudes and belief changes after using AI to perform four different work tasks (developing interview questions, fact checking, creating online content, and writing a recommendation letter). In Study 1, we find that while positive attitudes increase and negative attitudes decrease towards AI post-interaction, the latter effect is much stronger. However, our findings also reveal that people tend to wrongly predict their changes in negative attitudes after using AI, such that they expect that interaction with AI will exacerbate negative attitudes. Overall, we find that people are more willing to work with AI after interaction. Our second study finds further support for the relationships discovered in Study 1 and in addition, we find that changes to task-use perceptions (e.g., acceptability, performance, suitability, and willingness to use AI for a certain task) are highly dependent upon task types.
Additional Details:
The event is free and is open to the general public.
The direct link to the seminar will be sent out to registered participants 2 hours before the event. The JCB Bioethics Seminars are now being live-streamed to our YouTube channel. If you don't receive the link 2 hours before the event, please head over there and click on the live stream. Subscribe to our channel to receive notices of upcoming events.
Questions?
Please email Terry Yuen, jcb.ea@utoronto.ca.
Register Here
