Technical Fellowship
Every semester and summer, CAIAC runs an 8-week introductory reading group on AI safety, covering topics like neural network interpretability¹, learning from human feedback², goal misgeneralization in reinforcement learning agents³, and eliciting latent knowledge⁴. The fellowship meets weekly in small groups, with dinner provided and no additional work outside of meetings.
See here for the curriculum (subject to change).
Applications for the Spring 2025 Techincal Fellowship are open! Apply here.
Application Deadline: January 31.
For those interested in the policy and governance side of AI, we recommend applying to our introductory policy fellowship. It is possible to participate in both programs.
Past intro fellows have primarily been undergraduate, masters, and graduate students from Columbia. The fall and spring fellowships are strictly in person on campus, so we can't accept non-Columbia affiliates due to campus security. The summer session is remote, so students from other NYC universities and non-student professionals are welcome to apply.
Participants should be familiar with basic concepts in machine learning, such as deep neural networks, stochastic gradient descent, and reinforcement learning. See Week 0 of the curriculum for recommended introductory materials. We may group cohorts according to previous experience.
We ask for your availability in the application, and will attempt to accommodate people's schedules when forming cohorts. Each cohort meets once a week for two hours, with dinner or lunch provided. We'll be meeting in various meeting rooms across Columbia's campus depending on availability.
If you've already read all the material in the curriculum, you may be interested in attending our research groups. Feel free to email us at cualignment@gmail.com to discuss other ways of getting involved with CAIAC!
The fellowship is facilitated by CAIAC members with research experience in AI safety. This includes upperclassmen and graduate students.