The Intro to AI Safety Seminar is an 8 week course which explores the impact of increasingly powerful AI systems on society, explains the problems which the AI safety field is currently working on, and goes over cutting-edge reinforcement learning and deep learning safety research by OpenAI, DeepMind, and Anthropic. You can expect to learn from many prevailing sources in the field which will help you develop a solid research agenda that you may work towards in future semesters.
What is covered:
The seminar consists of a series of video and reading lessons which are accompanied by weekly, collective meetings that expand on the topics covered in the lessons.
We will adapt the curriculum from BlueDot Impact's Future of AI and Technical AI Safety courses, which cover topics like progress in AI, reinforcement learning from human feedback (RLHF), mechanistic interpretability, and AI control. We may also add more information on technical basics like the Transformer model.
How it works:
The seminar will last about 8 weeks, beginning on the week of February 9th and ending approximately the week of April 6th (with the week of March 16th skipped due to spring break). This is an estimate, as we are both testing a new curriculum this semester and experimenting with a more free-flowing discussion.
The seminars involves about 30 minutes of independent reading and 1 hour of discussion each week. The discussion will be led by senior club members and involve guided questions.
The time and place of the seminars will be based on the availabilities of everyone who fills out this form. We will try our best to ensure you are able to make our sessions but cannot make any guarantees.
Prerequisites:
We’re looking for students who meet any of the following requirements:- Curiosity about AI safety
- Interest in or familiarity with machine learning concepts
- Basic programming knowledge
However, if you do not meet the prerequisites or are not sure if you meet the prerequisites, feel free to still fill out this form or show up! Seminar meetings will be open to everyone.
Timelines:
The scheduling deadline is Sunday, February 1st, at 11:59 PM ET.
We will email out the seminar time by Friday, February 6th.
In the meantime, you can learn more about us at
https://aispurdue.org/, or contact us at aisafetypurdue@gmail.com.
We also have a Discord, which is linked at the end of this form.