AI use cases

AI safety research and alignment

Evaluate model behavior, train against reward hacking, and help align future AI systems.

From Ryan Greenblatt – What happens once AI can automate AI research? by Dwarkesh Patel