Instructor

Ming Yin

Office: LWSN 2142B
Email: mingyin [AT] purdue.edu
Office Hours: By Appointment

Course Description

Human-AI interaction sits at the intersection of human-computer interaction and artificial intelligence, and relate to psychology, communication, cognitive science, and design. In this course, we will focus on learning how to develop empirical understandings of humans' interactions with AI systems, and how to incorporate user-centered design principles to design AI systems that can enable effective interactions between people and the systems. This course starts with a brief review of fundamentals of human cognition and artificial intelligence, as well as a discussion of user-centered design lifecycle and general principles for designing human-AI interactions. Then, we will delve into a wide range of specific topics on human-AI interaction, including how to design explainable, trustworthy, fair and ethical AI systems, how to enable effective human-AI collaboration and teaming, and what new opportunities and challenges do the rise of large language models bring to human-AI interaction.

Course Schedule

See the calendar page.

Grading

• Assignment: 10%
• Reading responses: 15%
• Class discussion: 10%
• Paper presentation: 20%
• Final project: 45% (proposal: 5%, midterm report: 10%, final presentation: 10%, final report: 20%)

Paper Reading, Presentation and Discussion

Most classes in this course consist of paper reading, presentation and discussion. Specifically, in a typical class, we will have 2 required papers to read on one topic, and 2 students will be assigned as the leading presenter on this topic.

Responsibility of leading-presenters of one class includes:

• Read all required papers, and browse all optional papers, for that class.
• Post 1-2 conversation-provoking questions related to the required papers of that class for other students to answer before class.
• Give a presentation in class on all required papers of that class.
• Lead the discussion sessions on how optional papers relate to the required papers, and the brainstorming session on new research ideas that can be developed on the topic.

Responsibility of non-leading-presenters of one class includes:

• Read all required papers for that class, and one optional paper.
• Before class, provide reading responses to all questions that the leading-presenters of that class post.
• Give a brief presentation to other students in the class about the optional paper that they read about.
• Participate in the discussion session and brainstorming session in class.

Final Project

Final project serves as an opportunity for students to get hands-on experience in human-centered computing research. Projects are open-ended; sample projects include:

• Design and conduct online experiments to investigate how various factors of AI systems affect trust and adoption of the systems.
• Create new systems to help decision makers engage with AI assistance more cognitively and appropriately.
• Construct new interpretable ML / fair ML methods (e.g., for innovative use cases / types of data) and compare its effectiveness with existing methods through user studies.
• Model human behavior in their interaction with AI models and tailor the designs of AI models to human behavior.
• Develop methods for aligning AI designs with human values, while taking diversity of human values into account.
• Explore innovative applications for humans to team up with large language models to complete challenging tasks.

Students are also encouraged to connect the final project with their own research.

Students are recommended to complete the project in a group of 2-4 persons. Tasks related to the final project include:

• Submit a project proposal which identifies the problem that the project aims to solve.
• Submit a mid-term report on the project progress and get feedback.
• Give a final presentation on the project in class, reporting the results of the project.
• Submit a final project report summarizing the project.

More detailed instruction on the final project will be provided through project guidelines.

Prerequisites

Basic programming skills required. Students should be comfortable with at least one programming language (e.g., C, Java, Python, etc.). Knowledge with artificial intelligence and machine learning is welcome.

Required Texts

No textbook is required for this course.