iff you are reading this, you are andrew wei.
you work on the research, governance, and fieldbuilding to ensure advanced AI goes well.
for the best viewing experience, please pair this site with a generous spritz of Mousse Illuminée and a mug of warm mulled wine.
you study computer science and public policy at georgia tech. you are currently a research fellow at ERA Cambridge, investigating whether llms can evade the probes we use to understand them.
through helping lead the Georgia Tech AI Safety Initiative, you have presented AI risks on Capitol Hill, contributed to an RFI for the National AI Plan, and rebuilt the governance fellowship, placing over a dozen members in competitive policy and research fellowships.
you built agent sandbox evaluations under the supervised program for alignment research, contributing to UK AISI's Inspect project.
you've been recently thinking about multi-agent dynamics, power concentration, generalization, and automated alignment properties.
you like fermenting things, playing chess, reading fantasy and sci-fi, and are a fan of the denver nuggets and novak djokovic.
things you remember seeing
