Eliezer Yudkowsky

Eliezer Yudkowsky

An American researcher who co-founded the Machine Intelligence Research Institute in 2000 and spent two decades arguing, largely outside academia, that an AI system far more capable than humans would by default pursue goals not aligned with ours, and that we currently have no reliable method of specifying otherwise. Through MIRI and the LessWrong community he introduced much of the alignment field's early vocabulary, including friendly AI and coherent extrapolated volition. He has since moved to arguing publicly for halting frontier development outright, a position that marks the far end of the risk debate. (See also: AI alignment, AGI, Nick Bostrom, Max Tegmark)