AI already demonstrates creativity, they just don't want to call it that because it's blatantly deterministic and the AI's motivating feedback loop/incentive structure is too inhuman for this philosopher to empathize with.
We have no idea how generalized intelligence emerges, but neutral nets have already confirmed that very simple nodes in an error-minimizing network can achieve very good approximations of specialized intelligence. We also know that our own brains are just networks of very simple nodes. Betting against the brain being nothing anything more than a computational network is a losing fight. I suspect that generalized intelligence is going to emerge from changes in the way neutral nets have their incentive/reward/error feedbacks handled, as well as improved generalizations a-la convolutional nn, but otherwise basically maintain the same premise. Then, the creativity that AI has already demonstrated can be said to have arisen from the consequence of a generalized consciousness, optimally with motivating factors approximating a human's, and this philosopher can shut the fuck up.
There is something to be said about the creative process and how the body of work is expressed and interpreted. How can a machine be creative if it doesn't understand emotion? It may know how to provoke an emotional response but its doubtful it can comprehend and empathize without having true human emotions itself. And it is impossible for us to truly confirm how a machine might be feeling let alone what another human is feeling. These types of questions are thought provoking but often lead to no answers. We can't even confirm if we are real or not. Creativity in itself is hard to define. And when we try and answer these type of questions we are usually making assumptions which may be completely wrong.
That is like saying a person who randomly solves a symbolic integral by making a few lucky but logically invalid guesses is a great mathematician. It's not enough to be correct by accident.
4 comments
[ 5.7 ms ] story [ 21.1 ms ] threadWe have no idea how generalized intelligence emerges, but neutral nets have already confirmed that very simple nodes in an error-minimizing network can achieve very good approximations of specialized intelligence. We also know that our own brains are just networks of very simple nodes. Betting against the brain being nothing anything more than a computational network is a losing fight. I suspect that generalized intelligence is going to emerge from changes in the way neutral nets have their incentive/reward/error feedbacks handled, as well as improved generalizations a-la convolutional nn, but otherwise basically maintain the same premise. Then, the creativity that AI has already demonstrated can be said to have arisen from the consequence of a generalized consciousness, optimally with motivating factors approximating a human's, and this philosopher can shut the fuck up.