[CS575] AI Ethics / Spring 2024

All contents in this document are tentative.

Important Links for the lectures.

• Zoom Link
• Google Drive
  • All the lecture materials and assignments will be shared in the google drive.

Important Schedule about the lectures.

• [20240228] No offline class today; Instead, watch the recorded video.
• [20240304] No offline class today; Instead, watch the recorded video.
• [20240306] No Class ! Have a break :)

Announcement

• [20240311] Submit Google Forms about team sign-up. (Group) by 23:59 Mar 11(Mon).
• [20240226] Submit Lecture 1 Survey
• [20240226] Join slack channel: Invitation Link

Teaching Staff

• Lecturer: Alice Oh
• TA: Juhyun Oh (411juhyun@kaist.ac.kr), Eunsu Kim (kes0317@kaist.ac.kr)
• Contact: Through slack "qa" channel

Time & Location

• Mon/Wed 14:30 PM - 16:00 PM
• Rm. 2445, E3-1 (Information Science and Electronics Bldg.)

Prerequisites

• Knowledge of machine learning and deep learning (CS376, CS470, or CS570)

Schedule (Subject to Change)

#	Date	Topic	Reading	Presenter	Notes
1	2/26	Introduction to AI Ethics		Lecturer
2	2/28	Overview		Lecturer	Recorded Video
3	3/4	Generative AI		Lecturer	Recorded Video
	3/6	NO CLASS
4	3/11	Overview		Lecturer	Form teams & sign-up
5	3/13	Generative AI		Lecturer
6	3/18	Bias & Fairness		Lecturer
7	3/20	Mitigating Social Harms in LLMs		Invited Lecturer (Sachin Kumar)
8	3/25	Safety (toxicity, jailbreak)		Students
9	3/27	Safety (toxicity, jailbreak)		Students
10	4/1	Truthfulness (misinformation, hallucination, sycophancy)		Students
11	4/3	Truthfulness (misinformation, hallucination, sycophancy)		Students
	4/8	Project Proposal		All students	Recorded Video
	4/10	NO CLASS			General Election (국회의원 선거)
	4/15	NO CLASS			Midterm Exam Period
	4/17	NO CLASS			Midterm Exam Period
12	4/22	Privacy Issues in Data & Models		Students
13	4/24	Privacy Issues in Data & Models		Students
14	4/29	Transperacy & Limitation of Current Gen AI		Students
15	5/1	Explainable AI		Students
	5/6	NO CLASS			Substitute Holiday (어린이날)
	5/8	Project Progress		Students
16	5/13	Societal Impact & AI Divide		Students
17	5/20	AI for Social Good		Students
	5/22	NO CLASS			Substitute Holidays (부처님오신날)
18	5/27	Societal Impact & Environment		Students
19	5/29	Wrap-up		Lecturer
	6/3	Final Presentations		All students	In class
	6/5	Final Presentations		All students	In class
	6/10	NO CLASS			Final Exam Period
	6/12	NO CLASS			Final Exam Period

Course

This course includes lectures, readings, discussions, quizzes, and team projects. Students will be asked to do the following things.

Tasks	Descriptions
Project	Proposal, progress update, final presentation / Final report / Peer review / Teamwork report	1x	Team
Paper Presentation	30-minute presentation with 1 or 2 papers on a topic according to the schedule (will depend on the amount of content in the papers)	1x	Team
Discussion Prompt	Write 3 discussion prompts about a paper	2x	Individual
Discussion Presentation	Present the discussion of the paper based on their report	1x	Team
Paper Reading Reflections	Write reflections of the paper	2x	Individual

Lecture

Lecturer or student groups will give a lecture on each topic by each day.

Reading Reflections

Students will read, present, and think about the latest research from the reading list published in AI and ML conferences (e.g., NeurIPS, ICLR, ACL, CVPR, FAccT) related to ethical considerations. Readings may also include blog posts, articles in the media, online forum discussions, and publications from global governing bodies.

• Choose a paper related to the lecture topic from the reading list.
• Read the paper before the discussion and write a 1-page reflection on the paper, including a summary, strengths, limitations, and suggestions.

Discussion

Students will lead peers to discuss the readings with thought-provoking questions. You will challenge the findings in the articles as to their accurate reporting and interpretation; you will discuss relevance to the current time and various locales with different cultural backgrounds. You will present and discuss ideas for future research directions in AI and ethics.

• 20 in-class discussions (see schedule).
• Organize a group of 3-4 people, and have time to present what you read and discuss
• All groups should submit their result at the end of class.
• See the details on this page.

Team Project

Team project will be a major part of the class, especially during the second half. Projects will be basically replications or modifications of recent research in AI Ethics. See the details on this page.

Attendance and Participation Policy

• If you miss up to 2 classes, there will be no penalty. After 2, points will be taken off. Because you can miss up to 2 for free, we will not take any excuses for missing the class (unless you have a special case, such as prolonged sickness, in which case you should email the teaching staff).
• Unless otherwise specified, we will not accept late homework assignments, quizzes, peer evaluations, or project submissions. For exceptional individual circumstances, please contact the teaching staff.

Policy on Large Language Models

Recent progress in large-scale language models (LLM), such as ChatGPT, motivates explicit policies.

• The entire course policy is LLM-agnostic: no grader will ever evaluate your submission differently because they suspect it was generated by an LLM.
• You are free to use an LLM as long as you acknowledge it.
• Like any other online tool, you are ultimately responsible for whatever you submit.
• You will be asked to state how you are assisted by LLM at the end of the semester to evolve in future courses.

Evaluation (Subject to Change)

• Participation and Attendance: 20%
• In-class Discussion / Reading Reflections: 30%
• Project: 50%