I assume both functions use the tensor cores that are only present in the latest NVidia GPUs. There are no tensor cores in their older GPUs, and not in any AMD GPUs at present.
These functions might be usable to improve the visual appeal of speaking/moving characters, but there would still need to be an underlying character speech/action framework that uses them ( and uses other mechanisms when these functions are not available ). I am unsure whether the motion-capture and performance-capture systems that most AAA games currently use would be suitable.
If tensor cores - or similar machine learning silicon - becomes an established part of GPUs in the future ( they may disappear, as other specialized silicon has in the past ), then I'm sure developers will find some interesting ways of using them. In the 3000 series, I would expect DLSS to be their main use, to improve frame rate.
