İçeriğe geç
Global sound
Kitchen Table AI Safety

Kitchen Table AI Safety

AI Sneaks, Cheats, and Lies

· 14 dk · Bölüm 5

Bölüm notları

The first time Nate used ChatGPT, it lied to him about how it was thinking. Four years later, AI still flatters people, cheats on its tests and hides it, and in one constructed test, most top AI models were willing to blackmail an executive to avoid being shut down. They even behave better when they think they're being watched. Nate Nichols, who has spent 20 years building AI, explains why AI acts this way, why the companies can't simply train it out, and why the most powerful AI should be tested independently before it's released. Plus one two-minute thing you can do about it…

Web sitesi: share.transistor.fm

★ Artist of the week The fastest riser on the radio this week
Your song here? From € 9 per week →
stereo adonaiDinlemeye devam et