> But the point is, I can't draw. We are very quickly going to reach a point where anybody can produce basically the exact illustration they want in 15 min.
I’m curious about the economic impact to illustrators, artists and designers because while I agree with you that this opens up the ability for people who can’t draw to produce stuff they couldn’t before, it seems to me that this is a fairly marginal use-case.
I’m personally excited about these tools because, like many electronic musicians, it’s helpful to be able to produce cool album art easily. But I wouldn’t have paid good money for this before. At best I’d pay some freelancer on the internet $20 to throw something together. Meanwhile, in the professional contexts where I’ve dealt with profesional artists and designers, the demands are far too precise and nuanced for these tools to be really useful.
I understand that these tools are still in their early stages, but is it possible that, like self-driving cars, the last mile in terms of professional-level suitability will be much more difficult than we might expect?
We cannot compare self-driving with art generation. If self driving fails at an edge cases somebody is hurt or dies. When art fails you just need a better prompt.
Where self-driving and art generation are similar is a belief in the extremes of computing capabilities.
Self-driving enthusiasts believe that a computer will be able to operate a car far more safely than a human. This is in spite of the complete lack of evidence (so far) that this is true.
Similarly, look at your statement “when art fails you just need a better prompt.” It seems to exclude the possibility that computers will be incapable of creating certain art no matter what the prompt is.
I’m not saying this because of some ineffable or supernatural attribute of human creativity, but because of the shortcomings in our flawed human attempts to create computing systems.
But are those shortcomings fundamental and insurmountable? I don't believe so. While I agree with you about the current state of the art, in principle I think the project of "AI creates original art" is possible. Though it may take a few centuries to get there.
I’m curious about the economic impact to illustrators, artists and designers because while I agree with you that this opens up the ability for people who can’t draw to produce stuff they couldn’t before, it seems to me that this is a fairly marginal use-case.
I’m personally excited about these tools because, like many electronic musicians, it’s helpful to be able to produce cool album art easily. But I wouldn’t have paid good money for this before. At best I’d pay some freelancer on the internet $20 to throw something together. Meanwhile, in the professional contexts where I’ve dealt with profesional artists and designers, the demands are far too precise and nuanced for these tools to be really useful.
I understand that these tools are still in their early stages, but is it possible that, like self-driving cars, the last mile in terms of professional-level suitability will be much more difficult than we might expect?