Tuesday, November 29, 2022
2:00 - 4:00 PM (PST) ZOOM only!
Open to GSE faculty, students, and community.
Abstract:
Whether intended or not, we quickly render judgments of the people we encounter. From the sound of a voice alone, we make inferences about the social categories that may be applied to a speaker (e.g. gender, language status, etc.), their current emotional state (e.g. sad, angry), relevant traits they may possess (e.g., friendly, trustworthy, intelligent), and their potential attitude or stance toward a task, toward ourselves or others. As speakers, we also engage in behaviors to manage the impressions we make on others. In fact, there is some consensus over the importance of emphasizing one’s likability and competence, particularly during initial encounters (Cuddy, Fiske, Glick, 2008). This presentation will describe recent work to develop an automated tutoring system capable of providing users with dynamic and summary feedback on their impression management skills. The application is made possible through use of a flexible cloud-based pipeline that utilizes participants’ speech to infer social cues - reporting the extent to which participants would be perceived as likable and/or competent by a reference group. Definitions over the ‘reference group’ can be made flexible, providing an avenue for engaging communities in the automation process and community-determined inferences and feedback to users.
About the speaker:
Seth Corrigan’s formal training is in measurement with a focus on designing and developing digital learning applications, measures, psychometric modeling and machine learning. He has experience leading the design and development of traditional and automated assessment systems for national markets, and a successful track record of designing and developing measures for complex and challenging constructs. Examples of such measures include a range of soft skills, complex problem solving, oral communication skills, collaboration and written argumentation. Over the last decade he has designed and developed several digital games and simulations that utilize machine learning and psychometrics to simultaneously teach and assess, as well as making contributions to an AI driven embodied tutor for mathematics with automated gesture and speech. From 2014 – 2016 he was a member of the Embodied Design Research Laboratory at the University of California, Berkeley; and from 2016-2018 Seth was a visiting scholar at Indiana University’s Creativity Lab. He has also been a visiting researcher with New York University’s GovLab identifying means to support public participation in design of 21st Century skills and currently holds a position at UC Irvine’s Donald Bren School of Information and Computer Science. His work has been funded by the Bill and Melinda Gates Foundation, the National Science Foundation (NSF), the Hewlett Foundation, the Google Foundation, the U.S. National Oceanic and Atmospheric Administration (NOAA), as well as the Carnegie Corporation, among others. His most recent publication is a report for USAID on simulation and game-based technologies for supporting and measuring civic engagement in new democracies. His current work, funded by the National Science Foundation, investigates the automated detection and measurement of epistemic confidence using the acoustic and lexical content of speech.