According to LessWrong, there's a significant mismatch between what the AI safety community is perceived to be working on and what it actually is. Most people assume a large fraction focuses on alignment — ensuring that superintelligent AIs follow human values and instructions. In reality, only a handful of organizations like the Alignment Research Center and the newly announced Sequent focus directly on this. The rest of the AI safety effort is scattered across indirect work: capability evaluations, risk assessments, policy, and understanding misalignment itself. The authors note this without making a value judgment, but it's worth knowing: the field's effort is more distributed than most realize.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton