We make advanced AI systems safer and more transparent by developing methods to understand model behavior, model internals, AI agent interaction, and collective action in agent populations.
Contact us via galke@imada.sdu.dk