PERSON
Eliezer Yudkowsky
The autodidact who co-founded the field of AI alignment, spent twenty years warning that building minds smarter than our own is the last invention we get to supervise, and who has come to believe, with deliberate and grieving clarity, that we are very likely to get it catastrophically wrong.
Eliezer Yudkowsky is the most consequential thinker in the history of AI safety who holds no academic degree. Born in 1979 and self-educated from adolescence, he co-founded what became the Machine Intelligence Research Institute in 2000, not to warn about AI but to build it faster—convinced that a superintelligence would solve every other problem humanity faced. What changed him was not a loss of nerve but an act of thinking the problem through to its end. The closer he looked at the engineering of getting a superintelligent system to want what we want, the more impossible it appeared. He arrived at the
alignment problem—the challenge of building an artificial intelligence whose goals remain reliably beneficial as it becomes more capable than its creators—and concluded it was both unsolved and possibly unsolvable in the time available. Through LessWrong, which he founded in 2009, and the hundreds of essays