Introductory Courses in AI Safety

Applications for our intro programs are going to open soon. 

Express your interest here to be notified about the upcoming round!

These seminars serve as an introduction to alignment and governance to many, and is offered at other institutions such as Cambridge, Harvard, MIT, Oxford, McGill, and many others. 

Intro to AI Alignment 

For 8 weeks, you’ll spend 2 hours per week with a group of 2-8 participants and a discussion facilitator, switching between reading or watching material and discussing, there are no required readings outside of meetings. We aim to give you an overview of AI alignment — the research field aiming to align advanced AI systems with human intentions — and will touch on topics like interpreting the internals of neural networks, reinforcement learning from human feedback, and concrete ways that transformative AI might go really badly. Prior machine learning experience is not necessary. This session will be run in parallel with AI Safety Hungary chapter in Late July-August and you can have a look at the shared curriculum here.

Intro to AI Governance 

For 8 weeks, you’ll spend 1 hour a week with a group of 2-8 participants and a discussion facilitator, discussing the material you have spent around 1 hour reading outside of the meetings. We aim for participants to gain foundational knowledge for doing research or policy work on the governance of transformative AI. It covers research up to 2023 on why AI governance may be important, what large-scale risks AI may pose, which actors will be important in steering the trajectory of AI, and what strategic considerations and policy tools may influence these actors. This session will be run in parallel with AI Safety Hungary chapter in Late July-August and you can have a look at the shared curriculum here.