> The best possible next-word-predictor is omniscient.

Omniscient for the set of "meaning" embedded into it's training set. It's not broadly omniscient, big difference.

Right, but what's the limit of what you can deduce computationally from truly vast training sets? How much structure is there in the subtext of what's written down? It looks like there's rather a lot.

Sure, but same for any intelligence. The "training set" is just "the environment of adaptation".