Physicist-turned-AI-safety-researcher publishes resource guide: SPAR, MATS, PIBBSS and more
IAmTimNguyen · x · 2026-09-30
Karthik Viswanathan, a theoretical physics PhD now working in AI safety, published a resource list for physicists considering a similar move.
First research projects: SPAR (3-month part-time remote), MARS (early-career), Iliad Fellowship (3 months full-time applied math for alignment), PIBBSS (funded 3-month program designed for cross-discipline researchers, no prior AI safety expertise required), Pivotal (15 weeks, London), CBAI Fellowship (10 weeks, Cambridge, covering interpretability, multi-agent safety, formal verification), and MATS (selective, alignment/interpretability/security).
Taught programs: BlueDot's free intro courses and ARENA's hands-on training in deep learning, transformers and interpretability, with self-study materials online.
More from Companies & People
- OpenAI launches ChatGPT Space, taking on Microsoft and Google in productivity — firstadopter · 2026-09-30
- UK AISI says it found jailbreaks and alignment flaws at ~all major AI labs, is hiring — javirandor · 2026-09-30
- Dan Shipper: Sam Altman Talks Platforms Like Bill Gates Talked Microsoft — Create More Value Than You Capture — danshipper · 2026-09-30
- Dots Is Bringing Microsoft and OpenAI Back Together — TheTuringPost · 2026-09-30
- OpenAI Dev Day feels like WWDC of years past, with a live keynote instead of a pre-recorded ad — Dimillian · 2026-09-30
- OpenAI DevDay 2026: 20+ Launches, and the Five That Matter — Dots, Space, Decisions API — danshipper · 2026-09-30