I believe that video codecs will probably work along these lines at some point. Maybe have a base AI layer and then a more traditional codec layer on top to ref

I believe that video codecs will probably work along these lines at some point. Maybe have a base AI layer and then a more traditional codec layer on top to refine it. I think there are already AI-based codecs for reducing bandwidth of video calls with faces.