> did we inadvertently train AIs on idiotspeak.

Nope! That is - training on lowest-common-denominator, low-signal high-noise "idiotspeak" was not at all inadvertent.

In-advert-ently?