Generative AI and photography -- what will it mean for you?

[rdeloe]

Generative AI -- think ChatGPT for text and Stable Diffusion for images -- is going to have a huge impact on just about everything. It's making one side of my job easier (e.g., summarizing masses of information into a few core ideas), and it's making other parts so much harder (fighting plagiarism in student work).

Photographers are already using AI tools in Photoshop and other image editing software. You may think those tools are a good thing or a bad thing, but they're here to stay. Something else that is here to stay is photo-realistic images created from text prompts. Anyone with an Internet connection can make images that already look an awful lot like the kinds of photographs I see in this forum all the time. The quality is low now, but the pace of AI development is incredibly rapid.

What is this technology going to mean for photography in general, and for your photography in particular?

If you like thinking about these things, you might enjoy reading an article where I explore this topic. I'd be happy to chat about the ideas in this thread.

Generative AI

www.robdeloephotography.com

 

[buildbot]

Well written and excellent article, I think you nailed it. The human to human connection is what we can hold onto in the face of generative AI.

In fact, I am kind if excited to see where this pushes photography. When painting portraits and realistic scenes became less interesting to artists as photography became prevalent, what happened?
We got new, exciting kinds of painting and artistic expression! Now that anyone can generate a picture of aspen trees in the fall in Colorado, what we will explore instead?

 

[hcubell]

We got new, exciting kinds of painting and artistic expression! Now that anyone can generate a picture of aspen trees in the fall in Colorado, what we will explore instead?

Well, this was already true long before Generative AI. Nonetheless, while there were already tens of thousands of images of aspen trees in the fall in Colorado, there were only a handful that were truly wonderful and inspiring. This will always be true. There will be millions of horrible, cheesy images created with Generative AI, but there will still be very few creators of exceptional images using that technology.

 

[Ai_Print]

I'm not a fan of it and there are already issues like war images of Ukraine and Gaza on a couple of stock sites that are AI generated and not marked as such. World Press Photo just reversed a move to allow the AI category in it's journalism competition after some major blowback and the Washington Post did a great article on it regarding how it could quickly devalue the notion of photography being important and useful in the narrative.

A good friend of mine and famed wildlife photographer Paul Nicklen teamed with some people to come out with an app that will help greatly in preventing images that are to be uploaded to the web from being scraped for AI databases, it goes online soon:

Overlai | Protect Your Images From AI Theft

Safeguard your images and artwork from AI theft and protect your creative work with the Overlai app and plug-in. Explore our innovative solutions for the creator economy, including copyright protection, AI theft detection, and blockchain-based security features. Join Overlai to preserve the...

www.overlai.app

I know AI is in more and more of what we use in anything that is digital and in using it within a specific tool, it certainly has value. But I believe in the power of real talent and the power of a single real photograph to deeply affect it's viewers and I am not a fan of the notion that everything produced henceforth will be looked at as fake.

This is why I am now selling off some digital gear, will buy no more after this year and will reduce its use to just a few areas and concentrate more and more on film and darkroom based prints for the rest of my career. Those prints will not be online in public form and will be marketed in unique ways.

 

[rdeloe]

Well written and excellent article, I think you nailed it. The human to human connection is what we can hold onto in the face of generative AI.

In fact, I am kind if excited to see where this pushes photography. When painting portraits and realistic scenes became less interesting to artists as photography became prevalent, what happened?
We got new, exciting kinds of painting and artistic expression! Now that anyone can generate a picture of aspen trees in the fall in Colorado, what we will explore instead?

I'm glad you enjoyed the article. It's been in the works for a while as I've been thinking about what generative AI means for my own photography. We're each going to have to come to terms with this technology -- and those terms will be our own. A great example is @Ai_Print switching away from digital and back to film. I can see a lot of people pulling back to something solid and defensible.

One thing I've noticed about the world of photography and art is there are two very distinct groups of people, and for the most part they don't intersect. There are artists who use cameras and photographs, and then there are photographers. The former will see imagery from generative AI as a bonanza, and integrate the tools into their practice. They're already are quite unimpressed by my suggestion that a large part of what makes photography valuable and important is its connection to the real world. The latter group is typified by the people that haunt this forum. We're the ones who will typically see generative AI as a threat to the thing we love, rather than an opportunity.

I don't share your optimism, but I absolutely share your hope that generative AI will push "photographers" -- those of us who value the connection to the real -- into new creative realms. I hope that generative AI will make people re-evaluate what they do, and focus a lot more on making images that play to the strengths of the medium. That's the perspective I took in my article (and that's my own personal roadmap).

 

[rdeloe]

I'm not a fan of it and there are already issues like war images of Ukraine and Gaza on a couple of stock sites that are AI generated and not marked as such. World Press Photo just reversed a move to allow the AI category in it's journalism competition after some major blowback and the Washington Post did a great article on it regarding how it could quickly devalue the notion of photography being important and useful in the narrative.

A good friend of mine and famed wildlife photographer Paul Nicklen teamed with some people to come out with an app that will help greatly in preventing images that are to be uploaded to the web from being scraped for AI databases, it goes online soon:

Overlai | Protect Your Images From AI Theft

Safeguard your images and artwork from AI theft and protect your creative work with the Overlai app and plug-in. Explore our innovative solutions for the creator economy, including copyright protection, AI theft detection, and blockchain-based security features. Join Overlai to preserve the...

www.overlai.app

I know AI is in more and more of what we use in anything that is digital and in using it within a specific tool, it certainly has value. But I believe in the power of real talent and the power of a single real photograph to deeply affect it's viewers and I am not a fan of the notion that everything produced henceforth will be looked at as fake.

This is why I am now selling off some digital gear, will buy no more after this year and will reduce its use to just a few areas and concentrate more and more on film and darkroom based prints for the rest of my career. Those prints will not be online in public form and will be marketed in unique ways.

Your friend's app looks intriguing. I never post images from my phone though, so I'm hoping this kind of technology can be integrated into other tools. What would be ideal for me is if the protective technology is integrated into the SmugMug plugin that I use to post images directly from Lightroom to my website.

Collectively we're going to come up with a range of strategies to "defend" our work from AI.

• Some people are going to rely on the technological defense (like your friend's app). That's going to be an endless arms race.
• Other people are going to retreat into personal satisfaction. If you're already making images mostly for personal emotional and intellectual satisfaction, this is a fine strategy.
• I expect to see a lot of people doing what you're doing. I used to be very interested in medieval history, so I'll the analogy of pulling the defenders off the curtain wall around the castle and retreating to the heavily fortified keep for the final defense. You can hold out a long time in the keep if it's well fortified and has a water supply. As long as film, paper and chemistry is available, people using your strategy will be able to hold the line.

I'm curious to see what other strategies people come up with. This is not the first time we've been challenged by a new technology, but this one feels different to me. It's going to be a wild ride.

 

[rdeloe]

Well, this was already true long before Generative AI. Nonetheless, while there were already tens of thousands of images of aspen trees in the fall in Colorado, there were only a handful that were truly wonderful and inspiring. This will always be true. There will be millions of horrible, cheesy images created with Generative AI, but there will still be very few creators of exceptional images using that technology.

I hope you're right that generative AI will prove to be just one more new technology that we adapt to successfully. Only time will tell, but I think we're facing a difference of kind, rather than a difference of degree. I don't think generative AI is comparable to the "threat" photography offered to painting, or film offered to glass plates, or digital offered to film.

We shall see, and fortunately (or unfortunately?) we won't have to wait long to see how things turn out. The time scale for transformative change is now months rather than decades.

 

[cunim]

Why is film a defence against generative AI? Create an image on the computer, display it on a good monitor and snap a picture with your LF camera. Easy.

I agree that most of us here fall into @rdeloe's personal satisfaction category and will adapt or reject AI on that basis. Those who feel threatened by AI will soon have their revenge. Cliches will dominate AI, even more than they do photography. Very few AI practitioners will transcend the sameness and become competent professionals/artists. The ones that do create a niche deserve respect, because it will be hard to make a living at AI. Soon, there will be such a low entry barrier that hordes of newbies will be offering near-worthless product to a saturated market. In contrast, those of us in the personal satisfaction photo group will find ways to distinguish what we do from the generative stuff. After all, there are still people who paint for fun, and when was the last time an orchestra replaced a violinist with a synth?

Now, where did I put that Sinar?

 

[rdeloe]

Why is film a defence against generative AI? Create an image on the computer, display it on a good monitor and snap a picture with your LF camera. Easy.

I agree that most of us here fall into @rdeloe's personal satisfaction category and will adapt or reject AI on that basis. Those who feel threatened by AI will soon have their revenge. Cliches will dominate AI, even more than they do photography. Very few AI practitioners will transcend the sameness and become competent professionals/artists. The ones that do create a niche deserve respect, because it will be hard to make a living at AI. Soon, there will be such a low entry barrier that hordes of newbies will be offering near-worthless product to a saturated market. In contrast, those of us in the personal satisfaction photo group will find ways to distinguish what we do from the generative stuff. After all, there are still people who paint for fun, and when was the last time an orchestra replaced a violinist with a synth?

Now, where did I put that Sinar?

I think film is only a defence against generative AI if you do what @Ai_Print is proposing, in other words, make prints, keep them analogue and off the Internet, and market them (presumably) how we used to do it (perhaps face-to-face). That keeps them out of the training set (although realistically that won't make a huge difference). More importantly, I would say that making analogue prints and marketing them to people through interacting with them goes a long way to strengthening that human-to-human connection.

That all sounded good... but I'm reminded of that famous Mike Tyson quotation: “Everyone has a plan until they get punched in the mouth.”

 

[daz7]

AI development will affect the market more than the invention of a steam motor - the new industrial revolution is coming and it will decimate a lot of jobs making them obsolete.
But, did all the horses die because of the steam, electric and diesel engines?
The same will happen to art - most will be replaced by the cheaper and uglier AI but not all. The cheapest and nastiest creations of AI will supplant most of artwork created by humans, but the best photographs, paintings or drawings will still command a premium and will be exhibited, admired and sought after by connoisseurs.

 

[dj may]

I will not be using generative ai simply because it is plagiarism on an massive scale. I read a 600-page book on J.S. Bach; it was loaded with footnotes, endnotes and other references. You will find none of that with generative ai, yet the users are passing off the results as their own work.

Even worse, multi-national companies are supporting it even as they are quite litigious when they think their intellectual property has been improperly used.

 

[dchew]

I LIKE to think the images I produce and sell portray the emotion I feel when in the woods. Truth is, I am rarely successful. Regardless, I suspect the emotion, joy, curiosity, celebration or whatever you want to call it would be even more difficult to communicate if the experience was sitting in front of a computer. Specific emotions could, of course, be part of the instructions given to the AI generator. But often I can't describe those emotions in words. In fact, that is probably why I like photography so much; I'm not a writer for good reason!

If all we are producing is pretty pictures, I'm worried about our future. But if we are producing descriptions of our experiences, I'm excited.

Dave

 

[Ai_Print]

Why is film a defence against generative AI? Create an image on the computer, display it on a good monitor and snap a picture with your LF camera. Easy.

For me it is not a defense but merely where I was already headed in my post-commercial career. So it might actually be more of a validation that I made the right choice? To use a phrase by Rob above, I am already highly "fortified" in my stock of analog equipment, film, paper, chemistry and lab equipment, have been building it for years. The prospect of AI changing the overall market and public "value" of a photograph along digital lines makes this bath all the more soothing a choice.

As for the take a pic of an AI image on a screen with an analog camera, there will always be a way to sneak the system. I have no doubt that fine art photographers who use digital who are already well established in reputation will for the most part, be able to continue on in their chosen medium. But for the vast majority of those who are essentially "Dentist by Day, Fine Art Wildlife Photographer by Night" it will be game over in very short order.

But AI is a much bigger problem than photography, the phrase "Be a Man of Your Word" and why that even exists is gravely at risk. There are a lot of unhappy people in the world who become even more unhappy and angry when they are misguided and misinformed and god forbid, radicalized. Unless turned around in their tracks, they too will become even more nefarious and join the ranks of the misinformation brigade.

With a host of new tools tools in which to do that, I have great concern in how human to human trust can survive.

 

[AndereObjektiv]

Mean to me? Not a lot, honestly, I enjoy thinking, creating and doing for myself and having actual human friends. Mean for me, quite a lot I expect. Bruce Schneier has an article this week "AI and Trust" which provides a decent framework to begin to think about how AI and Chat/Prompt interactions can be problematic, to put it mildly. He's got another I haven't read about AI spying. The behavior of AI and the recent re-definition of lying to 'hallucinating' is typical of the newspeak generative grammar of modern AI language. It's going to be a big deal. Although I do appreciate the AI masking in Capture One and the AI signal processing of DxO PureRAW. So, like most innovations, plenty of good and bad.

 

[Ai_Print]

Generative AI -- think ChatGPT for text and Stable Diffusion for images -- is going to have a huge impact on just about everything. It's making one side of my job easier (e.g., summarizing masses of information into a few core ideas), and it's making other parts so much harder (fighting plagiarism in student work).

Photographers are already using AI tools in Photoshop and other image editing software. You may think those tools are a good thing or a bad thing, but they're here to stay. Something else that is here to stay is photo-realistic images created from text prompts. Anyone with an Internet connection can make images that already look an awful lot like the kinds of photographs I see in this forum all the time. The quality is low now, but the pace of AI development is incredibly rapid.

What is this technology going to mean for photography in general, and for your photography in particular?

If you like thinking about these things, you might enjoy reading an article where I explore this topic. I'd be happy to chat about the ideas in this thread.

Generative AI

www.robdeloephotography.com

In regards to your section of "Where to Go from here?"

• To avoid being part of the training set for generative AI systems, I think a lot about how I can make photographs that are distinctive and authentic. The photographs I need for research projects at work tend to be more conventional given the target audience, but even for those I'm striving for a distinctive voice that won't be easy to reproduce with text prompts in generative AI systems.
The people in tech know this and know that as it improves, that will not matter as long as there is access to your work online.

• I try to think about the specific people, or kinds of people, I’m trying to reach with my photographs. It’s much easier to build a human-to-human connection if you know the people to whom you are trying to connect.
This has always been a big part of my actual marketing and people will increasingly hunger for it. Human to human contact will solve a lot of the issues of provenance if you have all your technical ducks in a row, a great image not withstanding.

• In choosing subjects and themes, I ensure that it matters that I’m working with reality. If the image would work just as well with a subject created by generative AI, then I move on if I can .

In a way, I would personally suggest to avoid doing that, at least from a creative standpoint since Gen-AI is going to not discriminate either way. One reason is that it can disrupt the rhythm of your "jam" per se. When I am in the zone or about to be in the zone or even coming out of the zone, the whole image making process can be somewhat instrumental if not melodic and that even though the desired outcomes are great riffs, soaring solos and a memorable chorus visually speaking, the in between, chord changes and occasional dead note have to happen as well. In some ways, if you don't make the image you know is cliche, then the one you really want might not happen either.

• Finally, I think storytelling is an important way to make a human-to-human connection, so I work in groups of photographs as much as I can because it’s easier to tell a story with a group of photographs than with one. It helps that generative AI has a strong element of randomness. Currently, repeating the same prompt produces slightly different images every time. Thus, actual photography is still the most reliable way to create a series of images of specific subjects for a project. That may not be the case for long though if future developments make it easier to create a series of linked images with text prompts.

There are tech-heads working for Dall-E, Midjourney, etc. who read forums like this one who want nothing more than to disrupt and prove you wrong, that they can and will make Gen-AI able to everything and anything we say it can not. And they are going to succeed at it because they already are.

When I taught three semesters at this rather good photo program a few years back, the one thing I told the students on the last day with a very serious tone is that no matter what you want to do to in photography, help to keep it important. Don't just decide what voice you want to have as an image maker but also as a viewer of images.

 

[John Leathwick]

What a great discussion you've kicked off, Rob. My contribution will be similar to what others have already said,
in that my photography is very much focused on the things that I connect with. For a lot of my photos, even some
taken 50 years ago, I can still recall where I was standing relative to the subject, the light and the circumstances
around the photos - the image becomes a touch point to a particular event, mostly somewhere in the wilds or
in the garden at home, and collectively they act almost as a visual catalogue of events through life.

By contrast, I see AI generation as having no ability to create anything new or significant - as Noam Chomsky
commented recently, AI is ultimately just a very powerful tool for plagiarism based on its uber-extensive
sucking up of everything that it finds on-line. As such, I see AI generated images as basically a pastiche
extracted and blended from its catalogue of images, but such images are devoid of the touch point that
stimulate the capture of images as I practice it. In other words, it is not unrealistic to view it as a highly
technically accomplished exercise in meaninglessness - unless you have some other purpose such as
the propagation of particular perspectives or to satisfy your curiosity about what uber-computing can do.

As someone of an older generation (who has also delved deeply into the application of machine learning
in understanding biodiversity patterns), I'll be happily continuing to accumulate images that capture my
experience of life, and will share them with family and friends to communicate my fascination with the
natural world. I don't see any need to incorporate AI generated components from other peoples work
in order to achieve that.

John

 

[rdeloe]

In regards to your section of "Where to Go from here?"

• To avoid being part of the training set for generative AI systems, I think a lot about how I can make photographs that are distinctive and authentic. The photographs I need for research projects at work tend to be more conventional given the target audience, but even for those I'm striving for a distinctive voice that won't be easy to reproduce with text prompts in generative AI systems.
The people in tech know this and know that as it improves, that will not matter as long as there is access to your work online.

• I try to think about the specific people, or kinds of people, I’m trying to reach with my photographs. It’s much easier to build a human-to-human connection if you know the people to whom you are trying to connect.
This has always been a big part of my actual marketing and people will increasingly hunger for it. Human to human contact will solve a lot of the issues of provenance if you have all your technical ducks in a row, a great image not withstanding.

• In choosing subjects and themes, I ensure that it matters that I’m working with reality. If the image would work just as well with a subject created by generative AI, then I move on if I can .

In a way, I would personally suggest to avoid doing that, at least from a creative standpoint since Gen-AI is going to not discriminate either way. One reason is that it can disrupt the rhythm of your "jam" per se. When I am in the zone or about to be in the zone or even coming out of the zone, the whole image making process can be somewhat instrumental if not melodic and that even though the desired outcomes are great riffs, soaring solos and a memorable chorus visually speaking, the in between, chord changes and occasional dead note have to happen as well. In some ways, if you don't make the image you know is cliche, then the one you really want might not happen either.

I take your point (and agree wholeheartedly). Another, perhaps better, way to explain what I was trying to say is to refer to what I would have done before AI. I haven't always been successful at staying off the beaten path, but I try. I've never been interested in making yet another version of an image that lots of people have made before. When confronted with those scenes, I move on, unless I can see a new kind of image that I haven't seen before. That's what I meant with this point.

• Finally, I think storytelling is an important way to make a human-to-human connection, so I work in groups of photographs as much as I can because it’s easier to tell a story with a group of photographs than with one. It helps that generative AI has a strong element of randomness. Currently, repeating the same prompt produces slightly different images every time. Thus, actual photography is still the most reliable way to create a series of images of specific subjects for a project. That may not be the case for long though if future developments make it easier to create a series of linked images with text prompts.

There are tech-heads working for Dall-E, Midjourney, etc. who read forums like this one who want nothing more than to disrupt and prove you wrong, that they can and will make Gen-AI able to everything and anything we say it can not. And they are going to succeed at it because they already are.

Exactly. We're only seeing the tip of the iceberg. As a case in point, the reason the images I used in my article don't look like much of a threat is that Stable Diffusion is trained on very tiny images (512x512 pixel images for v1.5 and 768x768 pixel images for v2). As a result, it can't output high resolution images. This is what you can do with generative AI when you train it with high resolution images. https://www.telegraph.co.uk/news/2023/12/04/aitana-lopez-ai-influencer-social-media-artificial/

When I taught three semesters at this rather good photo program a few years back, the one thing I told the students on the last day with a very serious tone is that no matter what you want to do to in photography, help to keep it important. Don't just decide what voice you want to have as an image maker but also as a viewer of images.

That last sentence is the one that scares me the most, because I have no faith in the ability of my fellow humans (as a whole) to care enough to value authenticity and human creation. We have demonstrated time and again that we're willing to settle for low quality -- even outright forgeries -- to save money.

 

[rdeloe]

What a great discussion you've kicked off, Rob. My contribution will be similar to what others have already said,
in that my photography is very much focused on the things that I connect with. For a lot of my photos, even some
taken 50 years ago, I can still recall where I was standing relative to the subject, the light and the circumstances
around the photos - the image becomes a touch point to a particular event, mostly somewhere in the wilds or
in the garden at home, and collectively they act almost as a visual catalogue of events through life.

By contrast, I see AI generation as having no ability to create anything new or significant - as Noam Chomsky
commented recently, AI is ultimately just a very powerful tool for plagiarism based on its uber-extensive
sucking up of everything that it finds on-line. As such, I see AI generated images as basically a pastiche
extracted and blended from its catalogue of images, but such images are devoid of the touch point that
stimulate the capture of images as I practice it. In other words, it is not unrealistic to view it as a highly
technically accomplished exercise in meaninglessness - unless you have some other purpose such as
the propagation of particular perspectives or to satisfy your curiosity about what uber-computing can do.

As someone of an older generation (who has also delved deeply into the application of machine learning
in understanding biodiversity patterns), I'll be happily continuing to accumulate images that capture my
experience of life, and will share them with family and friends to communicate my fascination with the
natural world. I don't see any need to incorporate AI generated components from other peoples work
in order to achieve that.

John

Thanks John. In the end, I think that what you describe is the most likely "strategy". In other words, people who love making photographs will just continue loving to make photographs and will have to choose to not worry about whether anyone else cares. In some respects, we're already at that stage without generative AI. Billions of photographs are uploaded to the Internet every day, so anyone making photos and uploading them in the hopes that "if you build it, they will come" is not being realistic. We already need to believe in our work enough to carry on; a flood of photo-realistic generative AI images is just going to require more of that belief.

 

[tcdeveau]

FYI there was another similar thread started by another member that got moved.

My outlook may change but at the moment generative AI doesn’t mean much to me for my photography. These days I take photos to put on my walls. AI will never recreate what I felt at the time I was hitting the shutter button or the emotions the finished products bring me.

Was reading books to my 5 yo tonight for bed time and looked over on his wall and saw a picture I took of him as a baby and our cats hand is on the bouncer and it looks like the cat is rocking him. There’s another pic of my wife pregnant taken in the spot where our wedding pics were taken…and another of her in the hospital with him right after he was born. I have a whole other series with the M10M documenting our life during the pandemic that I need to put together at some point. AI will not replace that and will not stop me from continuing to do what I do. I also think I’m getting to a point in my life where im just tired of learning new things, but then again, maybe I’m just burned out at the moment from getting two post grad degrees that took 10 yrs in total to do.

I recently found out that one of my IQ4 Achro/Alpa images was accepted for exhibition by a local museum for an exhibition on Ga photographers. It’s not the Lourve or the Met, but I’m really excited about exhibiting this massive (for me) 36”x48” print - wasn’t going to happen with AI.

I can see maybe some use for me in editing just because I’m OCD and don’t have time to do what I want myself anymore, so I can see myself using using some generative AI fill tools at some point. I have a big cityscape pic of the Atlanta skyline, for example, and there was roadwork done on the roads and the tone of the pavement doesn’t match. If I could use generative fill to replace the road with a click of a button, that’d be great. That’s not a typical editing situation for me though.

I imagine a lot of the stuff we’re seeing with midjourney and what not right now will be like the early days of HDR and can see it dying out to a degree. Then again, maybe not. If ppl enjoy using it by all means go for it, but I’m going back to putting my head in the sand so to speak when it comes to developments like this for photography. I really don’t care, at least not now.

 

[Shashin]

I think we are having a reality deficit. Everything is becoming stylized and contorted that our sense of reality is being skewed. This can be in simply ways like body image (removing perceived defects) or in darker ways with conspiracy theories. AI will continue to exaggerate that trend. Already in photography, AI is creating fake images that are perceived as are real and real images are being evaluated as fake (AI). As we get more saturated in these works, we are simply going to either be unable to evaluate what is real and what is not: are we simply going to accept these as real or we are going to be in a perpetual state of skepticism where we will not trust anything. And when no one can agree on a reality, we are going to find a disillusion that will end in a paralysis of judgement or a simply an anarchy where fact is simply what you say it is. But the good news is someone will make money from it...

The other problem I see is going to be a skill deficit. Yes, AI can make works of art in a style of musicians, writers, artists, and photographers, but they are superficial. However, it gives the impression of being skillful without the burden of actually having skills. Yes, AI will take someone with average skills and polish the results to appear good, but the individual actually lacks the skill to do these things without the technology. So if AI polishes your writing, what do you learn about writing well. Where do you get the critical faculty then to evaluate that result and correct it? Camera technology is already pretty sophisticate and can produce technically constant images with little effort, but in and of itself, it does not result in compelling images. When the artist gives their agency up to a machine, then where does the skill building experience come from that moves that person forward to make more than a superficially "perfect" image. It is like we are doomed to endless Adam Sandler movies--good for a laugh, but little else.