Always available and pleasant to talk to, general-purpose LLM chatbots could erode public mental health by inflating users’ self-understanding and reducing their tolerance of effortful human interaction. Such erosion, like social media’s, is hard to fix. Diffuse and cumulative, it rarely triggers outrage or policy change before it is entrenched, and addressing it hurts chatbots’ engagement-based business models, inviting resistance. This talk asks: How might we, as professional HCI designers and researchers, act on these harms before it is too late?
This talk proposes a new way of thinking about the systems we build to address tech’s erosion of public mental health: seeing them as seeds of better business models, better policy, and better politics. I first make the case for this argument through two empirical studies, then demonstrate how this reframing can change the systems we design and our measure of design success. I close with a call for professional designers to embrace such thorny, ill-defined societal problems, problems that end users and communities cannot solve with software they build by prompting an LLM.
| 11:45am - 12:15pm: | Food and community socializing. |
| 12:15pm - 1:15pm: | Presentation with Q&A. Available hybrid via Zoom. |
| 1:30pm - 2:15pm: | Student meeting with speaker, held in the same location. |
Qian Yang is an assistant professor in Computing and Information Science at Cornell University. She is a designer and HCI design researcher, best known for her research profiling AI as a material for software design innovation, and for AI application designs that help users make measurably better high-stakes decisions. Find her group’s work at designAI.cis.cornell.edu.
The talk will include joint work with Dan Adler, Ned Cooper, Ibrahim Emara, Meir Friedenberg, Thomas Gilbert, Jose Guridi, Margaret Hagan, Angel Hwang, Steven Jackson, Sabine Junginger, Beth Kolko, Emma McGinty, Tony Wang, Talia Wise, Richmond Wong, and John Zimmerman. Her group’s work has been supported by the NIH, the NSF, Schmidt Sciences, Google, among others.