Allow the artist to interact with scene objects/characters as a director would. This requires what I used to refer to as "intelligent objects" back in the 80's. You would give directions and the object would figure out what it needed to do and how it would move.
So there would be a higher level of conversation with the 3D entities than "Set rotation keyframe."
Mocap requires a pretty significant amount of manual cleanup. The VFX industry is not a fan of Andy Serkis's bragging, to say the least. =P
I was extremely impressed with this motion matching presentation from GDC 2016. It blew folks away at the time. That said it hasn't really taken off everywhere AFAIK. Although it's definitely used in places (sports games!) https://www.gdcvault.com/play/1023280/Motion-Matching-and-Th...
I guess I like the idea of "directing". But in terms of pure practicality I'm very skeptical. What does the directing toolset look like assuming we don't have full blown AGI? That's tough. And I'm not super bullish on AI for this type of thing. Although I've been consistently surprised by ML capabilities over the last year. So in 3 years I'll probably say "oh yeah it was obvious that ML could do that..."
The key point is tackling the animation "problem" from a new vantage point, as opposed to having an interface based on numerical interpolation as the final solution.
But in that one, the player pushes a button and the character moves accordingly, picking natural movements and combining / merging them into something that looks pretty natural. I don't think that'll be useful in close-ups, but can be for e.g. crowds or wider angle shots.
One advantage of authoring cinematics as opposed to games is the lack of a requirement for real-time interaction. The cinematic sequence can take a few seconds and analyze/plan accordingly.