The Doomsday Invention

ai-safetysuperintelligenceexistential-risknick-bostromdeep-learning

Abstraction: Nick Bostrom's superintelligence thesis and AI existential risk debate

Key points:

Bostrom's "Superintelligence" (2014) argues an "intelligence explosion" — AI improving itself recursively — could make AGI more dangerous than nuclear weapons
Bostrom introduced "existential risk" as a formal philosophical concept in 2002; founded Future of Humanity Institute at Oxford
I.J. Good (1965) first formalized the intelligence explosion: "the first ultraintelligent machine is the last invention that man need ever make, provided the machine is docile enough"
Deep learning revolution (~2010) via cheap GPUs + big data shifted researchers' attitudes; 2015 Puerto Rico conference produced open letter calling for AI to be "robust and beneficial"
Google acquired DeepMind for ~$500M with conditions: no defense/espionage use and an ethics board — Demis Hassabis framed mission as "solve intelligence, then use it to solve everything else"
Geoffrey Hinton told Bostrom he didn't expect AGI before 2070, yet continued research because "the prospect of discovery is too sweet" — echoing Oppenheimer on the bomb

Connections: Nick Bostrom · Future Of Humanity Institute · Deepmind · AI Safety · Superintelligence · Existential Risk