We funded the second iteration of SERI MATS, a seminar and independent research program aimed at identifying and accelerating the professional development of junior AI safety researchers.
Warlop and Zhang both had experience working on multiple field-building programs for the Stanford Existential Risk Initiative, such as hackathons and summer research programs. They had run one iteration of SERI MATS before.
SERI MATS gives people an opportunity to conduct research under the guidance of an experienced researcher. Current mentors include Ethan Perez (Anthropic), Owain Evans (Future of Humanity Institute), Neel Nanda (Deepmind) and Daniel Kokotajlo (OpenAI).
Outcomes: Since we funded the program many of their alumni have gone on to publish original research in top conferences, join safety teams at AI labs, and start their own AI safety research non-profits.