

It doesn’t comprehend anything. That’s the damn point.
Though it WILL “understand” what you did by the algorithms that break down code turning the variable into another token. So all you’re really doing is cozting yourself more time and money in the slop machine.




Nahh, this seems exactly like something that would happen from advanced text predictors, which is exactly what ALL of these LLM “AI” are. The instructions are in English, no doubt. Of course there is a strong association with English words despite, “in Mexican” surely being somewhere in the “instructions”.
This is just more ELIZA effect in action… Fucking braindead execs/etc failing to understand that it is literally just advanced text prediction that’s fed through a TTS “AI” driven system, that’s surely given just as tenuouse of a connection to Spanish as the text predictor…
(I know your comment seems aware of that fact given the AGI slamming, I just had to vent in a way that describes why it’s such bullshit)