The cost to implement a model that can do P(Y|X) can unsurprisingly be much lower than something that can model P(X, Y) and that's not just because its an easie

The cost to implement a model that can do P(Y|X) can unsurprisingly be much lower than something that can model P(X, Y) and that's not just because its an easier problem but economics and what just works easily. But in the world of the future which is likely look at the "LLM of everything spoken/written" and given you can solve P(Y|X) P(X|Y) from a model that can do P(Y, X); LLMs could just win because of integration with everything.