Video wird geladen...
Video konnte nicht geladen werden
If an AI lab is kicking off an intelligence explosion, would it be in their interest to tell the President or not? Scott Alexander and Daniel Kokotajlo think yes. "We made the guess that early in 2027, the [leading AI] company would deliberately wake up the president. Strategically, [not... show more
53,186 Aufrufe • vor 1 Jahr •via X (Twitter)
8 Kommentare

Transcript: Apple Podcasts: Spotify: YouTube:

Meta AI researchers are fretting over the threat of Chinese AI, whose quality caught American firms, including OpenAI, by surprise.

@slatestarcodex @DKokotajlo All of this lives in the imaginary world where a powerful AGI trained mostly on human prediction is *not* advising on how to manipulate the President, nor manipulating its own "owners". Normies react skeptically to that prediction so it gets left out of scenarios.

@slatestarcodex @DKokotajlo You could have saved yourself the time❗️ The brilliant American engineer Suchir Balaji (major contributor to ChatGPT), who was murdered in his San Francisco apartment on November 22, 2024, explains in his final, unfinished essay why LLMs can never achieve AGI.

@slatestarcodex @DKokotajlo À Positively double negative argument

@slatestarcodex @DKokotajlo Brick wall background again... 🤔

@slatestarcodex @DKokotajlo Which president :)

@slatestarcodex @DKokotajlo and they will not know how or what exactly the AI is doing... but it gets the job done. the takeover will be gradual then sudden.
