Autodesk has been generating correct perspective for decades.  I bet your modern Architect wouldn't have a clue how to do it by hand.

Autodesk's Revit is the professional tool for true Architectural CAD drawing.

You input the building's coordinates effectively and presto the mathematics is converted to true perspective from any viewpoint you want.

Cut and paste sampling of a view in a general perspective like a street scene for a car will always sort of 'look right' but if you mathematically analyze it you'll find errors.

Even a single point perspective can be screwed.  [one I did years ago was set with 13 generated VPs and a few more I 'fudged'].

This is the sort of thing that can happen.... typically a single point is a single point... not a vague "somewhere about here" ...

To close it off, I'll say this:

I'd much rather have AI remove mild inconveniences or process large quantities of data in seconds way before letting it do creative tasks for me. Being creative is what makes us human. A world where you can create art by pressing a button is a world that's only inhabited by soulless husks, not human beings.

Thanks for listening. I'm out.

Once upon a time there were complaints that using effects and filters etc in Photoshop wasn't 'creative' art.

Machine creativity....

Photographs were not Art they were a click of a finger.

I don't use AI.  It took me decades just to start using layers.  Eg nothing in that LiteSTEP theme or Winamp skin was drawn with layers.  It was 'per-pixel', as in drawn/painted.

Heck.  I saw what was happening in Architecture... and thus building design.... urban design.... our living environment... thanks to CAD design.

Which is why I used a Tee square and set square for half a century instead.

Mechanical Architecture is painfully obvious.

Cheap...fast...repetitive.

However.

Full-blown GUI skinning is well over a quarter century old.  One of the first was a guy called Fly7e.  He reverse-engineered [hacked] Winamp's default look and before long Winamp was a 'skinnable'.

LoneRunnr created a text-code Shell replacement for Windows called LiteSTEP.  There were many others but LS was one of the most configurable.

One universal complaint over all that time.... it was becoming continuously evermore difficult/complicated to skin for Windows which led [partly] to a falling off of new blood to the sport.

Now, If the use of AI can attract more people to the craft that's a plus.

Even if the results may be a bit 'twee' the interest may inspire people to get creative in spite of AI. 

AeroPurple, you know I respect you, and your art has given me much happiness, but I have to disagree.

AI is a tool, just like your paint brushes. It's the user (artist) that controls the image on the canvas. The artist imagines some image in their mind, and then creates that image to share with others. Is that so different from the artist imaging some image in their mind and painting it on canvas with paint and brushes?

Using AI to create art is a process. It requires trial and error to get the image the artist had imagined to look the way he wants. Similar to traveling around the countryside to find the right backdrop for a painting. It is still human creativity even though the artist uses a different tool than paint brushes.

I'm from an advertising background where digital art and design have been a thing for quite some time. Using a computer to create art was new once, but became mainstream very quickly as artists adapted to this new tool.

The art you have created for WinCustomize falls into this category. You use a computer and graphics software to create. How is that any different from using a computer and AI to create art. The artist is still creating an original image using a new skill they acquired to manipulate and image to reflect their emotion.

There are plenty of artists that crank out landscape paintings, one after another, in order to earn a living. Many times they are using an algorithm to produce these. And then there are the artists that create original, well thought out, paintings. It depends on how much time the artist spend with their creation. Manipulating text prompts to create an image is a learned skill which takes many hours of practice. The amount of time they spend working on an image is relevant to the quality of the art. Even though the art may have been created using parts of existing images, it's still an original idea that the artist has imagined.

There is no easy answer to the question of "How original is AI art?" Time will answer this question.

It is different by a significant amount. When an artist draws something on a canvas, they have to keep in mind things like anatomy, perspective, shading, and many others. All Generative AI does is place pixels based on probability, e.g. "I'm 50% sure that the pixel to the right would be orange, and 40% sure that it would be black - I'm going to paint it orange because it's more likely to be there". There's no skill involved in this, just math.

The problem is that it should be used as a tool, but Generative AI does not work this way. When you make an AI wallpaper, for example, everything is "painted" for you, and all you can really do is adjust the brightness, perspective, and add a bit of text. So, in reality, the AI is the "artist" and you're the tool. You've just been played.

And it's the same thing as asking a chef to make a steak for you, and then when it's all cooked, you say that you made it simply because ordering the steak was your idea. That's not how it works.

Again, no. The only skill that goes into creating AI wallpapers is typing a few words, while an artist needs to be aware of many things, like, again, anatomy, perspective, shading, as well as dozens of other choices.

For instance, using a car to get somewhere instead of jogging is more efficient, sure, but you can't call yourself an athlete when you do that. That wouldn't fly. Creating AI-generated content is no different, and you shouldn't get away with calling yourself an artist when you use it.

At the very least, people drawing on a computer instead of on a canvas requires the same skills and expertise; AI generating art is more equivalent to typing something into Google - it's a skill, but it's not a creative one.

Something important that I should mention is that AI is the digital equivalent of a meat grinder - it's never anything more than a sum of its parts. You can't feed a piece of salmon into it expecting a pinapple to come out the other side. AI can't create anything that we haven't. The only way for it to create something novel is to teach that novel thing to it directly - it would never come up with that on its own.

More often than not, AI generation tools use tons of texts and images without the author's permission to produce output, by the way.

Finally, I'm in no way against technological progress, but I believe that new technology should be used responsibly. Like, I'm not against smartphones, I believe that they're a revolutionary technology that allowed millions of people around the world to communicate instantly when they otherwise couldn't. What I am against is corporations exploiting this new real estate by making their websites as addicting as possible so they can profit.

Same with AI, I believe that it could be used to advance humanity by many milestones - its ability to process gigabytes of data and spit out a meaningful result from what otherwise looks like nonsensical noise is impeccable. AI generation of text and images (and, unfortunately, now video as well), on the other hand, is incredibly harmful. It has allowed misinformation to spread faster than ever before. It has also put many talented artists, who I might add, are already some of the most underpaid and mistreated workers, out of their jobs because they've been replaced by a machine. AI generation also uses unacceptably large amounts of fresh water and energy to function - more energy than entire ccountries. The only winners in this situation are corporations and their overlords, who are already filthy rich.

So, when you support AI generation, you support a manifestation of greed, it couldn't be simpler than that.

If you're still not convinced that AI generation is a massive threat, I don't know what to tell you. But I'm leaving this conversation for my own sake, because convincing you about something this obvious is a waste of my time.

I personally will never use it, and never have.  There's nothing you can say that I don't already know.

AI is the inept's way of creating 'art', simply painting-by-numbers except more so. 

I was doing Life drawing aka nudes when I was 17. that's 53 years ago.  My entire working career has been graphic art and Architecture.

As for potential copyright issues... you won't find too many who have gone into bat more for the rights of the legitimate artist than me... even if it was google translate trying to tell someone in Russian he was being an arsehole .... and ending up with "My hovercraft is full of jellied eels..." [You might have to look that one up].  That's part of the role of Moderating submissions to these sites.

However.

When International copyright law is ambivalent about AI and it's use then so are we.

Throughout History, most new technology has been perceived as evil at first. Some even feared the telephone when it was invented. Some doubted its practical utility, seeing it as a mere novelty rather than a serious communication tool. The same was true when the Internet was first introduced. So, who among you can go a week without access? The Stardock server outage has been a huge reminder of how much we rely on it.

It's true that most technological advances provide the ability to do good and evil, depending on how, and by whom, it is used. AI is no different.

AI can do amazing things which could help Humanity survive the next 100 years, but it can also, in the wrong hands, destroy Humanity. This is not a new concept.

My point is, that we need to accept this new technology for what it is. Both good and evil. It's up to us to guide it's development towards improving our lives, and fighting against the evil.

I'm hopeful that AI can help us fight climate change, develop new drugs to fight disease, and overall improve the Human condition.

With that said, AI art is a very controversial subject. Everyone is entitled to their opinion, and a healthy debate between friends is always good.   

From the little I have tried, it seems that you have to have a very good idea/skill of writing prompts to get anything useable from AI.  What am I missing?

AI would certainly help skin Windows 11 better. I believe the OS has many elements to it that frustrate the skinning community and it would be nice to have a tool to alleviate some of these headaches. AI would be able to do a lot of tedious tasks automatically to allow skins to looks a lot better on Windows 11 and that would be a good thing for Windowblinds.

I think the ideas advanced here have been excellent. I just hope it doesn't end here.

I think templates of prompts could be added...like stops in Ps actions and after choosing color/shape/details or whatever, the "action" continues on to the next stop point in the algorithm.