DxO is using billions of RAW files to train its AI; it should focus more on training for portraits rather than birds, landscapes, or architecture

Every small bird looks virtually the same, every mountain range remains unchanging, and every building is renovated only once every few centuries; in contrast, every portrait subject is unique—with skin tones varying drastically between Asians, Europeans, and Africans. Moreover, birds, mountains, and buildings do not pay photographers; only humans provide payment for their services.

What do you think works better on small birds or mountains than human, AI side ?

What AI training is about birds, mountain, architecture and not or worse for humans ?

Are you talking about masks ?

To those not interested in birds. I’d say people, of all races, are far more similar than all small birds.

But as @JoPoV asks, what exactly are you wanting AI to achieve?