Embedded Files
Join a 7-week reading group exploring state-of-the-art techniques in technical AI alignment.
Weekly 2-hour meetings on UBC campus with food provided
Small groups of ~6-8 people
No additional work outside of meetings
Interested? Apply here.
Who can apply?
We aim to accept primarily undergraduate and graduate students at UBC, but students from other schools or working professionals are encouraged to apply.
What are the dates?
TBD.
What will I learn during this course?
Most of our material is drawn from the AI Safety Atlas, a textbook on AI safety. The course primarily focuses on preventing catastrophic risks arising from future AI systems. We focus on catastrophic risks because we think they are neglected relative to their scale.
What topics will we cover?
Here is the current syllabus:
Week 1: Foundations
How modern AI actually works: neural networks, transformers, and how LLMs are trained. What "capabilities" means and how we measure them.
Week 2: Scaling and Forecasting
Scaling laws, benchmark prediction, and inference-time compute. Building your own model of how fast progress is likely to go and what could bottleneck it.
Week 3 : The Alignment Problem
Why capable systems don't inherit our goals by default. Specification gaming, mesa-optimization, and the gap between what we train for and what we get.
Week 4: Training Methods and Their Limits
RLHF, constitutional approaches, and where they break down. How does alignment faking work in frontier models?
Week 5: Evaluations and Catastrophic Risk
Dangerous capability evals, red-teaming, and ASL-style thresholds. Misuse, misalignment, and how complex systems fail.
Week 6: Current Affairs and Next Steps
Where the field stands right now: regulation, compute governance, lab commitments, and open research problems. Plus how to get involved.
When and where does each cohort meet?
We ask for your availability in the application and will try to accommodate your schedule when forming cohorts. Each cohort meets once a week for two hours in-person at UBC. The exact location is to be determined.
How much experience with machine learning do I need?
No background in machine learning is required. Those with more relevant technical skills will be given priority, but we encourage everyone to apply. We may sort cohorts by experience level.
Should I apply if I'm more interested in AI governance and policy?
Check out our AI Policy Reading Group. It's possible to be in this course and the policy group at the same time.
What if I already know all the material in the curriculum?
Please contact us at the bottom of this page to explore other ways of getting involved!
Who will be my facilitator?
TBD.
Google Sites
Report abuse