Which Graduate Programs Will Best Set Me Up for a Career in AI Safety?

By Jason Zeng @ 2025-02-19T13:22 (+4)

Hello! I'm a recent graduate from the University of Michigan with a Bachelor's in CS & Physics and minors in Math and Philosophy. I'm pursuing a graduate degree in Computer Science to develop my technical AI safety research skills, since my career goal is to contribute to AI alignment at a leading AI safety research company.

I recently received my first admission to a graduate program: Columbia's Master's in Computer Science! However, they've given me a short time to respond, which I was able to extend to March 10 (originally February 24 – is that not abnormally early?). Still, that feels early, as many other programs won't release decisions until April.

To prepare, I feel the need to rank my programs now so I can make timely decisions. If I accept Columbia's offer but later renege, I will lose a $4000 deposit. I can sustain this loss, but I'd prefer to avoid it.

My main question: How should I rank these programs based on how well they will set me up for a career in AI safety?

Programs I've applied to (in tentative ranking order):

• Berkeley | MS→PhD in CS
• Stanford | MS in CS
• MIT | PhD in EECS
• Columbia | MS in CS (admitted)
• UCLA | MS in CS
• UC San Diego | MS in CS
• Cambridge | MPhil in Machine Learning and Machine Intelligence
• Oxford | MSc in Advanced CS
• University of Michigan | MS in CSE

Some key considerations:

• Career opportunities in AI Safety, particularly in industry (e.g., Anthropic, DeepMind, OpenAI) or alignment-focused research orgs.
• Research opportunities and faculty alignment with EA and AI Safety.
• Professional network and credibility in AI research.

Notes:

• I ranked Berkeley and Stanford most highly based on my understanding of their flourishing AI safety communities.
• I conducted research on the faculty interests that are most relevant to AI safety - I've decided to omit these details in this initial post for concision. I'm wondering whether I should weight this information less significantly, especially for Master's programs.

I’d appreciate any insights from people familiar with these programs, AI Safety career trajectories, or general decision-making frameworks for high-impact careers.

Thanks in advance!