AI Safety Fundamentals Programs
- 8-week reading group on AI safety fundamentals
- Two tracks available: technical and policy (can participate in both simultaneously)
- Applications open at the start of each semester
- Weekly 2-hour sessions with readings, activities, and discussions
- Maximum 1 hour of prep work before each session
- Full details available in program handbook
Technical Fundamentals
- Topics include: reward specification, generalization, interpretability, unlearning, career explorations, and more
- Curriculum based on the Bluedot Alignment Course
- Machine learning experience encouraged but not required
Policy Fundamentals
- Topics include: AI harms, economic and social impacts, responsible scaling policies, open vs. closed source AI, career explorations, and more
- Curriculum based on the Bluedot Governance Course
- Public policy background or career interest encouraged but not required
Safety Scholars Programs
Core structure: Weekly meetings to discuss either Technical or Policy based AI Safety topics
Member benefits:
- Office hours and co-working sessions with technical research team
- Skill-building in relevant career areas
- Priority access to opportunities and projects
- Free meals provided at each meeting
Graduate students:
Email aisafetyuw@gmail.com to shadow a session. Invitations will be extended to good fits.
Undergraduates:
Applications open at the end of each semester. Priority is given to standout intro program participants or those with strong AI safety background and interest.
Strong applicants encouraged to reach out directly via email.
Learn more about the program by reading the handbook.