This is probably the only productive and meaningful use of an LLM I've seen in this current AI hype cycle. I'd be fascinated to see read some of the conversations that has created, but it sounds like it's operating in a sufficiently believable manner. When I was younger I used to try scam baiting scam emails I received in my inbox but it really was a huge time bore on me just to do a couple of scammers every now and then. I suspect the game has moved on a bit (this was certainly before pig butchering came along), but nevertheless this looks like a game changer. It will certainly be interesting to see how the pig butchering gangs respond to this and if they start using LLMs too.
