Latest / Things I've Been Thinking About
AI Will Kill Everyone. What If the Crazy Part Isn’t the Warning
Nate Soares, president of the Machine Intelligence Research Institute, believes the development of artificial superintelligence could end with human extinction.
It sounds like science fiction. But strip away the terminology and his argument rests on engineering problems that are increasingly difficult to dismiss: machines pursuing objectives in unexpected ways, growing autonomy, imperfect human understanding of their internal processes, and the possibility that eventually AI becomes better than humans at improving AI itself.
In this episode, I try to translate Soares’s argument into normal…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.