Pragmatisation is the Way Forward

Stephen Elliott argues in his latest LessWrong essay that the AI safety community should stop living in an academic bubble and start working directly with political and financial power structures. He says the field, which aims to prevent harmful outcomes from advanced artificial intelligence, has relied too heavily on pure intellectual norms and therefore lacks influence over the massive capital and regulatory forces shaping AI development. Elliott proposes splitting the discipline into three arms—intellectual, political and capital—and giving the intellectual side authority to guide the others toward concrete, enforceable safeguards. He warns that without such real‑world engagement, safety research may be sidelined as the race for advanced AI accelerates.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton