Proposal: pause pretraining to master control of model drives

A thought experiment proposes globally pausing major pretraining and RL advances to focus on controlling training-induced internal drives, with debate centering on how to quantify controllable alignment and how long scientific consensus might take.

2026-08-28 ~ 2026-08-28 · 2 related posts

• Thought experiment: Pause pretraining to focus on controlling inner drives — louisvarge · 2026-08-28
• Opinion: Pause Pretraining to Focus on Controlling Inner Model Drives — louisvarge · 2026-08-28