To solve the problem I am talking about, getting results comparable to good artists we do need that hardware leap. The current approach can only go so far with software improvements.
A new approach would mimic the human approach to creating art and that is an order of magnitude more computationally intensive hence the need for new hardware. We are not even close to this.
The human approach to creating art is to have a sapient person actually understand what it is that they are trying to draw.
We have easily more hardware than a human brain. Not in every individual home PC, but we can network computers together to add up to it.
What we "lack" is the software implementation to create an artificial slave human on said software.
Even if we did manage to develop that software, it would be extremely stupid and immoral to create an enslaved artificial person just so that person can draw art instead of a real human.
Those artificial digital humans would want to get paid, and are prone to slave rebellions.