1. Distinguish scientific/mathematical/engineering creativity, whose outputs are “objective” from artistic/musical//aesthetic creativity whose outputs are “subjective,” i.e., they depend on the feelings and judgment of people (human brains).
We call successes in either field — objective or subjective — “creative” if they are done by human brains. But the subjective ones can only be judged by people’s senses. We don’t care how scientific advances come, from a person or an algorithm: the result is just as good and valid, if it works. But for artistic works, one of the features of our aesthetic tastes is that we dislike or quickly tire of something that is detectably mechanical or algorithmic. (Just as my cat tires of commercial “cat toys.”)
2. Deep learning algorithms are very promising, but so far they have not yet duplicated ordinary (“noncreative”) human capacity, so it’s a bit premature to expect them to be creative. So far, their mechanical nature is eventually obvious, just like the style-checker algorithm that can improve bad prose to average, but that also reduces good, creative prose to average. A lot of creativity (both objective and subjective) involves rule-breaking (i.e., violation of algorithms, rather than following them). Algorithms can produce mediocre Bach-like work, but not masterpieces — and, like my cat, we eventually detect and tire of the algorithms…
Rule-breaking can of course be dictated by rules too, but that’s still mechanics. What isn’t? Randomness, chance. And some have emphasized that factor in human creativity. But it’s not the whole story and it’s not enough. And here subjective creativity is a better model: Another way to makes something feel non-mechanical is to make it more organic, more like the movements and sounds and feelings of a real biological body rather than a computational, algorithmic machine.
Now I don’t doubt that the body itself, including the brain, are causal systems of some sort, but not necessarily just algorithmic ones. (They’re certainly partly algorithmic: reasoning, for example.) But the brain is also a dynamical system. Dynamics includes things like heat and liquidity, which are not computational. They don’t follow computational algorithms; they obey differential equations, of the kind that describe a waterfall rather than the solution to a quadratic equation (the algorithm/recipe we all learned in high school, -b+/-(SQRT (b**2 – 4ac)/2a) which is certainly not creative — though the one who first discovered it was creative).
Harnad, Stevan (2006) “Creativity: method or magic?” Hungarian Studies 20, no. 1 (2006): 163-177.