Topics in Foundations of ML: AI Alignment and Safety
COMPSCI 2881R
Subject & Catalog Number
Course Information
Description
This will be a graduate level course on challenges in alignment and safety of artificial intelligence. We will consider both technical aspects as well as questions on societal and other impact on the field. This is a fast-moving area and it will be a fast-moving course. I will expect students to be able to pick up technical knowledge on their own. In a sense, the programming language for this course will be English: students will be allowed and encouraged to use AI tools for all homework and assignments. On the other hand, this means that expectations will be raised: it may well be the case that I would expect you to do assignments in a week that in previous years would have taken a month.
Course Notes
This course was previously numbered CS 229BR. Enrollment priority is given to PhD students.
Class Notes
Students who are interested in the course should complete this form: https://forms.gle/qHiNMeaFp83hnQ1HA
Recommended Prep
Students are expected to be proficient in programming as well as mathematically mature. You should be comfortable with concepts in statistics, optimizations, as well as programming and applied machine learning such as training and fine tuning a neural network. In particular we expect students to be familiar with all concepts taught at Harvard CS 1810 (formerly CS 181) or MIT 6.390. Most of all, you should be prepared to pick up both applied skills and mathematical topics on your own. I will publish a "Homework Zero" on my homepage before the course starts. Students will be required to complete and submit it in order to apply to be enrolled in the course.
Available for Harvard Cross Registration
