This is important to read, and listen. “The core observation is that when AIs goals conflict with humans goals, weird shit starts to happens “This is a legitimately huge problem, obviously scary to people outside the field.”

→ View original post on X — @paulroetzer