The output may seem rational, but was the process of synthesizing it 100% rational? Can you write a program that creates the Mona Lisa while explaining each decision along the way?
Yes, they are. I think we can agree that there is a beautiful elegance to it that is similar to that of mathematics.
But that doesn't answer my question, which asked how to write a program that was capable of outputting the Mona Lisa while rationalizing the decisions it is making. Shouldn't we be able to feed everything we know about the Mona Lisa into a program so it could generate something similar? If not, why not?
I'd say that simply because something is rational doesn't mean it can be codified into a computer program (yet). (Maybe my example code disrupted my point more than helped.) The design of most computer programs is rational, but a computer isn't sophisticated enough to program itself.
I'm actually having trouble thinking of an important human endeavor this was not in some way rational. Maybe some `Outsider Art` created by mental patients? Programmers who work in `BrainFuck` or `LOLCODE`? I guess I'd only think of something as irrational if I thought a reasonable person would not come to a given conclusion based on the same inputs (such as knowledge, skill, experience, etc).
Explain to me exactly what it is about creating the Mona Lisa that cannot be done by a machine, and I will write a program which does it (with apologies to Von Neumann).
In other words, this sounds like an "irrationalism of the gaps" theory.
Not because it cannot be quantified, but because you failed to quantify it. A sense of aesthetics is not an irreducible "thing," it's a set of algorithms and heuristics. If you could explain exactly why you find DaVinci's creativity so appealing, you still wouldn't be able to reproduce it directly--but you would have reduced it to a problem in NP.