roon argues AI alignment will be solved with scalable, bitter-lesson-style methods

roon (@tszzl) says the plan has always been to solve AI alignment directly rather than wait for human enhancement.

In a November 2, 2026 post on X, roon argues alignment should follow the bitter-lesson approach that has driven other progress in modern AI, meaning methods that scale with computation and data instead of hand-built rules. The post gives no specific technique, timeline, or evidence.