Fall 2025: Graduate Seminar in Machine Learning (COMP 640 004)

Responsible AI for Society: Foundations, Incentives, and Policy

• Time: Wednesdays 2:00 – 3:15 PM
• Location: HRZ 211

Overview: This seminar explores the theoretical foundations and interdisciplinary challenges involved in building responsible AI systems and algorithms. Topics include data privacy, fairness, incentive-aware algorithm design, strategic data sharing, and the legal and policy dimensions of AI, such as intellectual property and regulatory frameworks. By bridging ideas from machine learning and economics, the course aims to equip students with a critical understanding of the technical and societal implications of AI systems.

Syllabus: Syllabus PDF

Sign-up form: Sign up here to present (need to log in with the Rice account)!

Class Schedule:

Date	Topic	Material
08/27/2025	Introduction to Privacy	In-class lecture
09/03/2025	Introduction to Game Theory and Mechanism Design	In-class lecture
09/10/2025	Data Markets & Privacy (I): Data Acquisition Mechanism	– Selling Privacy at Auction – Optimal and Differentially Private Data Acquisition: Central and Local Mechanisms
09/17/2025	Data Markets & Privacy (II): The Role of Data Externality	– Too Much Data: Prices and Inefficiencies in Data Markets
09/24/2025	Data Markets & Privacy (III): Data Monetization & Regulations	– On Three-Layer Data Markets – Data, Privacy Laws and Firm Production: Evidence From the GDPR
10/01/2025	Data Valuation	– Towards Efficient Data Valuation Based on the Shapley Value – Data Shapley in One Training Run
10/08/2025	Fairness: An Overview of Definitions	– Equality of Opportunity in Supervised Learning – Inherent Trade-Offs in the Fair Determination of Risk Scores
10/15/2025	Fairness-Accuracy Trade-offs	– Algorithm Design: A Fairness-Accuracy Frontier – The Statistical Fairness-Accuracy Frontier
10/22/2025	Fair Resource Allocation (I)	– Best of Both Worlds: Ex Ante and Ex Post Fairness in Resource Allocation
10/29/2025	Fair Resource Allocation (II)	– Fair Allocation in Dynamic Mechanism Design
11/05/2025	Strategic Manipulation in ML	– Strategic Classification
11/12/2025	Human vs. AI in Decision Making	– Predict Responsibly: Improving Fairness & Accuracy by Learning to Defer – The Value of Context: Human versus Black Box Evaluators
11/19/2025	Copyright for Generative AI	– On Provable Copyright Protection for Generative Models
12/03/2025	Learning Preferences From Data	– Recovering Preferences From Finite Data – A General Framework for Estimating Preferences Using Response Time Data