I wonder if there's a way to adapt this to AI / machine learning. To my uneducated eye, there's a distinct similarity between the "breaking and reforming" of [s

I wonder if there's a way to adapt this to AI / machine learning. To my uneducated eye, there's a distinct similarity between the "breaking and reforming" of [simulated] physical bonds, and the training of a neural net. If you consider your training data (or other "external sensors") as the environmental forcing factors... do you process "information" instead of "energy"?