Great idea but they seem to just reimplement the same old ANN with photonics, instead of rethinking the concept of “learning” via reinforcement (or delta rul

Great idea but they seem to just reimplement the same old ANN with photonics, instead of rethinking the concept of “learning” via reinforcement (or delta rule, gradient descent etc.) which is why it wastes so much energy in the first place. That’s not how learning of objects and places work in energy- and time-constrained living organisms. AI will not come from engineering lab but from science lab (neuroscience in particular).