OpenAI Engineer Says Rationalist AI Program Won't Work

Keller Jordan, a research engineer at OpenAI who created the Muon optimizer, posted on X that he had been learning about the rationalist AI program. He concluded it will not work. The post includes images of archived pages from early writings by Eliezer Yudkowsky and discussions on LessWrong. The message stands as his direct statement without further elaboration in the visible post.