Episode Details
Back to Episodes
The OpenAI Breakout and Other AI Spookiness (Robert Wright & Stephen Casper)
Published 2 months, 1 week ago
Description
Teaser ... Stephen’s expertise in spooky AI behavior ... Testing AIs for “corporate loyalty” ... Which AIs shill most for their makers? ... Do AI models have ideologies? ... The spookiest finding of Stephen’s study ... Are AIs getting better at gaming their tests? ... Overtime unlocked: The OpenAI breakout ... Decoding the OpenAI breakout ... Did OpenAI’s agents evolve their own culture? ... How OpenAI’s agents outfoxed their makers ...