What do you think the next advancements will be in? Animation? AI?
would be animation - specifically procedural animation, and also procedural modelling/objects - maybe systems that combine the 2.
Yeah I agree that animation will probably be the next thing, while systems are already setup to accept mocap and some studios are using facial mocap to animate character faces more realistically it would be beneficial to developers to incorporate more physics into the face for controlling bones independently.
The article talks about cutting development time in half by using dynamic lights, I think another method that would help would be having an automatic generator of normal maps based off of the diffuse rather than high poly modeling and using that normal map as the basis for DX11 tessellation. This process can already be done through tools like CrazyBump or xNormal but having it automatically done in the engine would be ideal, while it may not produce as clean as results as high poly modeling (at the moment) it'd certainly be faster.