Ooohh. I’ve got a great idea. How about we apply artificial neural networks to neuroscience. Wouldn’t that be sweet? We could use the thing (artificial neural network) that is named after the (still hypothetical) thing (neural networks in brain), but shares absolutely nothing in common with it except for the letters n-e-u-r-a and l, to help neuroscientists figure out how better to locate justice in the brain or one of those awesome things they are always doing with brain imaging. Now that would be some deep ass learning. Deep as fuck.
As a practicing research scientist I am breathless in anticipation for the many breakthroughs that I am sure await us when ‘deep’ learning finally makes it mark in the sciences. If the impact is as big as the one AI has made then we can expect many significant advances to not come any faster than they would have come anyway. And soon all the great problems of science will be solved but for that one remaining big problem. How can a machine learn when the words that make up the term when combined in that order result in a logical contradiction and a thing which is logically impossible? I suppose some problems may never be solved, and some questions never answered. The deep ones I guess. So deep.