As AI advances, doomers warn the superintelligence apocalypse is nigh : NPR
The anthropogenic website on a laptop organized in New Hyde Park, New York, on August 22. Anthropic is one of the main artificial intelligence companies. The CEO of the company was one of those who signed a public declaration in 2023 recognizing the “risk of EA extinction”.
Gabby Jones / Bloomberg via Getty Images
hide
tilting legend
Gabby Jones / Bloomberg via Getty Images
Nate Soares, co-author of the book If someone builds him, everyone diessaid time is exhausted to prevent superhuman AI from wiping humanity.
Martin Kaste / NPR
hide
tilting legend
Martin Kaste / NPR
What happens when we make an artificial intelligence that is smarter than us? Some IA researchers have long warned that the moment will mean the fate of humanity.
Now that AI is advancing quickly, some “doomers” say it’s time to brake. They say that the automatic learning revolution that has led to a daily AI such as Chatgpt also made it more difficult to understand how to “align” artificial intelligence with our interests, which means a greater risk for humans once AI can surpass them. AI security researchers say there is a chance – a “p (doom)” where the “p” means “probability” – that such superhuman intelligence would act quickly to erase us.
Martin Kaste of NPR reports tensions in Silicon Valley on IA security.
For a more detailed discussion of the arguments for – and against – doom, please listen to this special episode of NPR explains:
AI and the probability of destiny
And for the really curious, a reading list:
The abridged version of the “everyone dies” argument in the Atlantic.
Refutation of “useful idiots”, also in the Atlantic
The potential calendar of an AI takeover
Research on “Faking” and deception
Smith College Economics professor James Miller is thinking about the theory of waiting for the Apocalypse AI, while hoping for the salvation of the AA
Maybe AI does not accelerate in smarter AI, at least not yet. Metr search
Analysis – and skepticism – Experts on the short -term probability of artificial intelligence of human or superhuman level
