Stanford CS547 HCI Seminar | Autumn 2026 | Designing AI for Societal Good

Item: Stanford CS547 HCI Seminar | Autumn 2026 | Designing AI for Societal Good
Rating: 4
Author: PostSingularity Institute

by Qian Yang · View the channel on PSI

Humanities, Social Sciences & Thought AI & Society UBDigital and Information technology: general topics UBJDigital and information technologies: social and ethical aspects

🎙 Qian Yang 👥 1.3M 📅 October 23, 2026 ⏱ 56 min 👁 37 📄 original study 🧭 2026-10-23

Available in: English (current) Français

Keywords

societal impactgenerative AIharm measurementpublic opinionpolicy change

Summary

In this Stanford CS547 HCI Seminar, Qian Yang, an assistant professor at Cornell, presents her lab’s research on designing AI for societal good. She argues that current approaches—building harm-free AI applications and calling for policy enforcement—are insufficient to address the pervasive, slow-burning impacts of generative AI on society, such as mental health and creativity. Drawing on historical precedents from HCI (e.g., smartphone overuse research) and a CHI workshop, she finds that impactful interventions often involve building harm measurement systems that catalyze public opinion and policy change over longer timescales. However, for generative AI, such measurement itself requires AI and funding, creating a deadlock. She proposes a new design philosophy: ‘designing AI as planting seeds’—where designers build systems that measure harm and seed changes in business models, policy, and politics. She illustrates this with ongoing design work. The talk emphasizes proactive, pragmatic steps designers can take despite incomplete understanding of societal impacts, aiming to break the current impasse and move from mere suspicion of harm to actionable change.

167 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk offers a compelling and original argument that reframes the typical ‘build a better app and call for policy’ approach. The speaker provides strong historical evidence from HCI (smartphone overuse, civic monitoring) to support her claim that harm measurement systems are often more effective catalysts for change than harm-free redesigns. She logically deconstructs the deadlock created by the cost of AI-based measurement and the prerequisite of public opinion for policy funding. The argumentation is coherent and well-structured, building from empirical findings to a practical design agenda. The value lies in shifting designers’ attention from idealistic but powerless artifacts to strategic, systemic interventions that operate on longer timescales. She acknowledges uncertainties and the need for multiple pathways, and her suggestions are concrete and actionable for system builders.

Scientific Rigor, Source Quality, Title Accuracy

The presentation is scientifically rigorous: it synthesizes literature from STS and HCI, draws on a CHI workshop with over 80 international researchers, and includes a thought summit with domain experts. However, specific citations for the historical cases (e.g., smartphone overuse research) are not explicitly named in the transcript, and the talk relies more on anecdotal evidence and expert consensus than on published metrics. The title accurately reflects the content and scope, and the description provides links to Stanford’s graduate programs and the HCI seminar schedule, which serve as general references rather than direct sources for the claims. The talk itself does not overstate its conclusions and explicitly notes the difficulty of proving causal societal impact, which enhances credibility.

260 words

Title / Content Match

The title accurately reflects the content: a seminar lecture on designing AI for societal good, with concrete case studies and a proposed design approach.

Quality & Reliability

8/10

The talk presents well-reasoned empirical research and design practice, grounded in a literature review and expert discussions. It acknowledges limitations and complexities, and proposes a novel framework for AI design. The speaker is a professor with relevant expertise.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

Stanford HCI Seminar Schedule — Official seminar series page where this talk is hosted
Stanford Online Graduate Education — General Stanford graduate programs info as linked in description

Concurring Sources

Stanford HCI Seminar Schedule — Institutional context for the seminar; no direct contradiction.

Contribution & Novelties

The talk contributes a novel conceptual framework for AI designers: rather than building ‘harm-free’ applications, they should build systems that measure harm and actively seed changes in public opinion, policy, and business models. This shifts the focus from individual user experience to systemic societal impact, and offers a pragmatic path forward despite measurement challenges. The historical analysis of HCI’s successful policy interventions provides a concrete playbook, and the identification of the cost deadlock for AI-based harm measurement is a fresh insight that has not been widely discussed.

Pour aller plus loin :

Science and Technology Studies — STS provides theoretical background on how technologies influence society and policy, which underpins the speaker’s analysis.
Human–Computer Interaction — The field in which the talk is situated, with a history of policy-relevant research.
Artificial Intelligence Ethics — Related to the responsible design of AI systems and their societal impacts.

146 words

Radar Profile

The radar profile shows strong scores in information quantity and quality, with moderate technical difficulty. The balanced profile indicates a substantive academic talk that is accessible yet intellectually rigorous, with slightly lower scores in technical depth reflecting its focus on design and policy rather than algorithmic details.

Reliability 8/10