What is the state of the art in natural language understanding?
I would say IBM Watson is the leader of the pack. However, Watson was not made in a vacuum; there was quite a bit of other research, not to mention government sponsorship, leading up to it. Of course, Watson was a triumph of brute force computing, which means to me the challenge now is how to make it more economical and accessible to everyman. Also, Watson clearly lacked voice input, so big opportunities for Nuance. Two of the core technologies leveraged by Watson were machine learning and hypothesis ranking.
See my 2011 blog post:
· How Many PlayStations Make A Watson?