Too many AI researchers think real-world problems are not relevant

I promise you if machine learning can be used to teach a robot to fold and put away my laundry you'll get some fucking impact all right.

Maybe they think they are doing basic research and not applied research?