There are fundamental theoretical issues with natural language parsing. Once we can grok how humans do it, then we at least have a model to go implement on another piece of machinery. Its pretty funny that for all the meaning we exchange in language- its hard to get the nuts and bolts of how exactly we do it.

Funny how things like balance, object recognition, perception of a context and simple intuitive extrapolations of the above have turned out to be some of the hardest technical challenges- while a few trillion floating points per second with pristine accuracy is easy.

heck, even acrobatic movement (teathered for stability) has turned out easier in some ways than picking up blocks and putting them on top of eachother.

Davin L.