ostinslife

If you define "deceive" as something language models cannot do, then sure, it can't do that.

It seems like thats putting the cart before the horse. Algorithmic or stochastic; deception is still deception.