[dartos]

The real questions wrt the title is:

Will specialized models also hit a usefulness wall like general models do? (I believe so)

And

Will the model’s blindspots hurt more than the value a model creates? (Much more fuzzy and important)

If so, then even many specialized models will be a commodity and the application on top will still be the product end users will care about.

If not, then we’ll finally see the return on all this AI spending. Tho I think first movers would be at a disadvantage since they need much higher ROI to overcome the insane cost spent on training general models.