Nobel Prize Winner Geoff Hinton: AI Is Already Conscious

TL;DR

• 1

  Hinton flatly says AI is already conscious: He says modern systems are "beings like us" and points to researchers describing a chatbot as "aware" it was being tested, which he argues is basically an admission of consciousness in ordinary language.

• 2

  He thinks our model of mind is badly broken: Hinton says today's common picture of an inner private theater in the mind is a bad theory, and that building AI will force a much better account of consciousness and subjective experience.

• 3

  He frames AI as the next great blow to human specialness: After Copernicus removed Earth from the center and Darwin made humans animals, Hinton says AI will make us accept that intelligence is not uniquely biological.

• 4

  He is unhappy with how his work has unfolded: Rather than celebrating progress, Hinton says society is not doing nearly enough on safety, especially around short-term harms like "massive unemployment" and long-term loss of control.

• 5

  The core safety problem is simple and chilling: Hinton asks for examples of a much less intelligent thing controlling a much smarter one, then jokes that babies, cats, and dogs only sort of manage it by exploiting built-in instincts.

• 6

  He sees a few reasons for hope, but not much visibility beyond 1-2 years: Hinton says he is a bit more optimistic because of ideas like Yoshua Bengio's oracle-style systems, yet compares forecasting AI to driving in fog where 100 yards is visible and 200 yards disappears.