When people quote their “conversations” with AI bots, I notice that the bots have a broad vocabulary of niceness and sycophantic obedience. The bots can say “I’m sorry” in dozens of creative ways.
A few years ago I was dealing with customer service chatbots that repeatedly scheduled mythical technician appointments, then apologized when the mythical technician didn’t show up. Their vocabulary was less sophisticated than the current versions.
WHERE ARE THEY SOURCING THIS TYPE OF TALK?
They sure as hell aren’t getting it online. No human ever says “You’re right and I’m wrong”, let alone apologizes. It happens occasionally in real life. Even if the AI thieves are monitoring cellphone audio these instances are so rare that they’d be hard to pull out of the noise of universal nastiness.
Polistra's Mill 