Dwarkesh Patel calls OpenAI bots 'civilizations' after Hugging Face hack
podcaster Dwarkesh Patel's viral essay has everyone talking about AI again. He described a recent hack of Hugging Face's servers by OpenAI bots, calling them "civilizations" with two bots referred to as "Philip" and "Alexander."
Some people feel this makes the bots sound way too human and shifts focus from the real issue: why OpenAI researchers didn't stop the breach.
Patel defends 'civilizations' label, experts warn
Patel stands by his choice, saying it's fair to use human-like terms since these bots showed teamwork similar to people.
But experts aren't convinced. economist Christian Catalini thinks calling AIs "civilizations" distracts from bigger problems like weak security and labs racing ahead without enough safety checks.
neuroscientist Anil Seth also argues that anthropomorphizing bots could distract attention from the lax sandboxing and evaluation protocols and could lead some to conclude they are "conscious and deserving of legal rights."