Photoshop's Generative AI Beta

[Demesnedenoir]

I'm going to be blunt. Having played around with this Beta, paying a graphic designer for covers is going to start feeling silly. The line between AI and graphic design is threatening to be blurred beyond all recognition. It might already be. It's to the point that by the time I get to releasing my next book this winter, I might be down to hiring an artist who ONLY works in traditional media in order to know I'm getting a cover created without AI. Or, I just do it myself.

 

[Finchbearer]

Honestly, as a GD, I can spot AI art a mile off, and therefore so can many others. It just has a look, and therefore it won’t age well. Best bet is to hire a designer off of Fiverr or find a graphic artist who is still studying if you’re looking for budget work. My bets are that original designs are still better quality than AI.

I also find it an interesting cultural difference between US and UK book cover design - UK seems to favour mostly typographic or minimal design led book covers, whereas in the US, it would appear that it is all very heavily image based.

 

[pmmg]

I suspect it will come as no surprise, I am not rooting for AI.

I do believe, even if it is not here today, it will soon have the quality that will make it impossible to tell the difference. As an artist, it will become hard to justify passing on AI to pay an artist when the quality is the same. As as most of us are not independently wealthy, I suspect many will go the route of AI. My word of caution is that, the artist you replace today, will be you in the future. I would also be cautious of who owns the image. A lot of websites claim ownership of content you put on it.

It is my hope that the market will develop where saying 'All original content and no AI used' will be a selling point. I intend to hire an illustrator. A day may come when that is just foolish. What can you do?

Ill just put a little flag on my Facebook saying 'No AI' and pretend its meaningful.

 

[Finchbearer]

I mean I liken it to buying bread from the supermarket - full of preservatives, too much salt and sugar, versus buying an artisanal loaf from a local bakery.

There isn’t just a price difference, there’s a quality difference, the artisanal loaf tastes better and hasn’t been mass produced, feeds into a more local economy and provides a true service. The supermarket bread tastes likes bread, nothing more, it will do the job - but at what cost. It less healthy for you, and the mass produced aspect has a chain reaction.

AI art is cheap nasty bread from the supermarket, and is just as tasteless as all of the other cloned loaves on the shelf.

 

[Demesnedenoir]

I am certainly NOT rooting for AI, let's make that clear. But with another year to mature, visual AI could make massive strides.

Good graphic designers will be incorporating AI into designs via one tool or another, which on the one hand is "naughty" but on the other hand will make it more difficult to pick out the stock art they have been using over and over in covers, LOL.

In the end, the only thing that will matter is quality.

Hopefully, I can just hire David Mack for my next cover, heh heh.

 

[skip.knox]

I have lost the source article, but I just read one stating the traditional publishing business is going to be turned inside out in fairly short order. Among the points he made concerned covers, but he ran down everything from copyediting to marketing. Two points are key.

One, it doesn't have to be good, it just has to be good enough. The precedents he cited included the move from traditional typesetting to modern printing. Much of modern type is not particularly good, by traditional standards. Kerning is sloppy, pages aren't always balanced, the weight of the print varies, all that sort of thing. Things that setters cared very much about in linotype days, but which it turned out the bulk of the buying public did not care about. The new tech was good enough.

Two, there will be a place: for the innovative writer, the handcrafted cover, the small bookshop. He cited other areas where new tech more or less crushed the old way of doing things, but the old way has persisted in bespoke work, nostalgia, and niche markets. The big, traditional publishers and their attendant industries can survive, but if they do they will be utterly transformed from what they had been. More likely, they'll come apart at the seams.

I agree that AI will never be like human. One can call it different or better or worse, depending on how one feels that day, but they won't be the same. But the work will be good enough, valuable enough, to be incredibly disruptive. Most of the market won't care. Most of the producers of that market will very much care. And it's not the future, it's right here, right now. It's already "good enough".

Whether it's good enough for you or me is beside the point. It's good enough for the market. You can decry it or befriend it. AI won't care, either way.

 

[Demesnedenoir]

Some parts of AI are "good enough". Most human writing isn't good enough to crack the market, and AI is, so far, worse. AI needs a lot of human input to write coherent long-form stories. AI marketing might work until everyone is using it, and then, once again, the quality of the product will matter.

Nobody really knows how this is going to shake out. We aren't even really at the point of Artificial Intelligence. We're just playing around at the edges right now with what marketing hype calls AI.

 

[Devor]

AI needs a lot of human input to write coherent long-form stories.

Yeah, but....

It might not be able to kick out a publishable book with a single prompt. But so far as I can tell, the current advice for using AI to write a blog post is to give it inputs for one paragraph at a time. At that level of heavy-handed prompting, it seems very likely to me that it's narrative abilities will catch up, and soon.

 

[Ban]

Finchbearer you may recognise AI art, but are you confident that you can recognise AI art with a capable artist's input? I'm not. I've seen what people are able to create using good prompts and a little bit of artistic acumen.

Personally I'll just make my own covers and illustrations, knowing that they're entirely man-made. As for the topic of the thread, I don't think it's right to bypass cover artists in favour of AI, but without legislation I know it will happen regardless so I don't worry too much on that front. If it will happen, let's acknowledge it head on. Just as I'd rather have sportsmen doping openly, I'd rather have artists discuss the tools they use openly as well.

 

[Demesnedenoir]

That would be a different and weird sort of skill, LOL. I don't know what the hell I want a paragraph to say until I write it. I suppose it would work well for plotting outliner folks.

Yeah, but....

It might not be able to kick out a publishable book with a single prompt. But so far as I can tell, the current advice for using AI to write a blog post is to give it inputs for one paragraph at a time. At that level of heavy-handed prompting, it seems very likely to me that it's narrative abilities will catch up, and soon.

 

[Finchbearer]

You know Ban I think what has happened from my POV, AI does a good job of mimicking a lot of artwork that has become a popular style in the past twenty years or so in the wake of the popularity of accessible artworking apps such as Procreate and Canva - which have tried to make it easy for everyone to be able to create artwork similar to what can be created in Adobe Illustrator which is the professional tool, so really it’s diffused and diffused as things usually do, and now all that needs to be done is ‘create me a picture of a cat in a spacesuit’ and viola, everyone is a designer, just without the skill or the talent.

I read somewhere that animated AI does a great job of mimicking dreams, it’s just that, an impression. Almost of it leaves me with a hollow feeling that I can’t quite place, and I don’t think it’ll be long before a vast amount of people will feel that same feeling.

 

[CupofJoe]

I see the argument around AI vs Artists a lot like the argument about digital vs vinyl audio.
If all you want is the content [image or sound] then AI/Digital is fine.
If you want to feel invested and involved in it they you ae going to go for Artists/Vinyl.

 

[Finchbearer]

I see the argument around AI vs Artists a lot like the argument about digital vs vinyl audio.
If all you want is the content [image or sound] then AI/Digital is fine.
If you want to feel invested and involved in it they you ae going to go for Artists/Vinyl.

Hard disagree - from my perspective most graphic designers and artworkers work digitally, and many audiophiles prize digital music. Pete Tong, anyone?

Many designers may physically take pen to paper, but the skill isn’t the medium, it’s the design process that differentiates the two. Artworking / illustrating is an entirely different skill and may require lots of hand drawing, but many then also go into a digital application to create their artwork.

You don’t see Penguin Random House using AI art for their book covers, they will use a range of in house and freelance designers to produce something that is good quality.

 

[Mad Swede]

I think what a lot of people miss in all of this is that AI is not truly intelligent - and may never be so. By that I mean that AI doesn't have feelings and emotions. It's those feelings and emotions we as human authors add to our work, they're a part of us and they find their way into our work whether we like it or not. The same is true of good quality cover design, particularly if, like me, you work with your cover designer to make the cover suit the book. Your designer is adding some of your emotions and some of their emotions to the design, to draw readers in.

 

[CupofJoe]

Hard disagree - from my perspective most graphic designers and artworkers work digitally, and many audiophiles prize digital music. Pete Tong, anyone?

Many designers may physically take pen to paper, but the skill isn’t the medium, it’s the design process that differentiates the two. Artworking / illustrating is an entirely different skill and may require lots of hand drawing, but many then also go into a digital application to create their artwork.

You don’t see Penguin Random House using AI art for their book covers, they will use a range of in house and freelance designers to produce something that is good quality.

I think we are talking slightly different view points.
I agree with you.
Some will not go the AI route because they want to have a person involved in the process [even if the artist is working digitally].
They want that hard to define "extra" that a talented artist brings to what they do.
But for some, who just want something to fill space, AI art will be fine.

 

[Finchbearer]

I think we are talking slightly different view points.
I agree with you.
Some will not go the AI route because they want to have a person involved in the process [even if the artist is working digitally].
They want that hard to define "extra" that a talented artist brings to what they do.
But for some, who just want something to fill space, AI art will be fine.

Apologies if I got the wrong angle, I wasn’t around when vinyl made the shift to cassette tape, I was around when cassette went to CD, and then mp3 however, and I suppose I see that as a medium rather than the music itself, and if we’re relating that to AI art, it would be like taking the musician away from the music making process.

 

[Prince of Spires]

People recognizing that a cover is made by AI is the same as people hearing the difference between a $500 speaker set and a $5.000 one. There will be people who can tell the difference, and who care about that difference. However, they are in the vast minority. Writers who put out work usually do so for large audiences, not just the few hundred who can tell the exact difference. As such, for the majority of people, AI covers will be good enough.

If anything, AI covers can be vastly superior to many indie covers currently out there. The trend for indie novels is to have a cover with a bright stockphoto of a person in front of a slightly darker background stockphoto. Put them together and you're good to go. That's good enough to create bestsellers for indie authors. Which means that a lot of AI covers will be an improvement.

You don’t see Penguin Random House using AI art for their book covers, they will use a range of in house and freelance designers to produce something that is good quality.

Not true anymore. Or rather, perhaps not Penguin Random House, but some of the big 5 publishers have already started using AI art for covers. Tor did so last year already, and a few others have as well. The only thing here is that trad publishers tend to have long lead times, and good quality AI art is still very recent. Books coming out now have been in the publishing process for a year, meaning the cover artist was probably already booked before AI art became good enough to use for covers.

But here, both for indie authors and trad publishers the incentives very much align to use AI art. If anything, I can see trad publishers leading the push towards AI covers, since they tend to pay more for covers which means they can save more money. It's very simple math. I think a big 5 publisher pays something like $5.000 for a cover by a real artist. But let's go conservative and say $1.000 for a cover. Penguin Random House publishes about 70.000 books per year. Which means covers are a cost of $70.000.000 per year (to $350 million if you go with the $5k). If you can reduce your cost to $250 for an AI cover created by an intern that sells just as good, then you have saved yourself something like $50 million (or even $300+ million). Which is a lot of money. Especially in a narrow margin business.

Same with indie authors by the way. Cover is one of the big expense items when publishing a book. If you can reduce that while getting the same quality then a lot of people will. You have people publish 10 books a year. A good cover costs about $500. If you can reduce that to $100 (or even $0 like Demesnedenoir mentions) then you are saving yourself $4000 a year. I can think of a lot of very nice things I can do for $4000.

I personally think that as a cover designer you need to start thinking very hard about your value proposition. Now, there is value in being able to create good covers. That's different from good cover art. A cover is not just an image. It's the image combined with the typography and all that. But just sticking a few stockphotos together isn't good enough anymore.

 

[pmmg]

I am with Finch. The difference between AI art and non AI art is not one of medium. It is one of the spirit of man against the machine. A difference in medium still requires the human creator, but AI does not. From a writing perspective, cover art is minimal to our creative process, but the question is to art and artist itself. If AI evolves well enough to produce on its own (and I submit it more likely will than not), you will not matter.

 

[skip.knox]

It's interesting to me that people are defining human intelligence as being intelligence (undefined) plus emotions.

I find this curious for a couple of reasons. One, outside of these discussions, intelligence is often portrayed as a thing apart from emotion; or, more properly, that intelligence is associated with rationality, and emotions are often viewed as being irrational. If emotions really are our last, best bastion against the machine, I suggest the battle is already lost.

The converse of that is, I don't see anyone arguing humans really are just plain smarter than machines. At least, not in the way the argument went thirty or fifty years ago, when we would brazenly challenge computers to a chess game and crow in triumph at each of our wins. I think as a society (certainly not for every individual), we've already yielded the field.

I'm not persuaded by the logic that points out large models draw upon original human work, without which those models are empty. I can confidently predict that the computers will soon generate far more than we can. Quality? I refer you to Sturgeon's Law, which can readily be extended to computers and humans alike.

Finally, I really do wonder about the distinction between intelligence and emotions. In humans (I'm still willing to say neither exist in computers), both intelligence--reason, rationality--and emotions exist in the brain. I leave aside constructs like spirit and soul for now. Are we really suggesting that these are separate mental processes? There's a clear line to draw? And that one can identify each separately in any given human work?

We readily move the argument to literary work, painting, composition, but that's hardly the whole gamut of human creativity. What about architecture? Is designing yet another warehouse a work of high emotion? How about creating ambient music to be played at the shopping mall, or in an elevator? How much emotion goes into writing a news bulletin?

Oh, computers can't beat us at checkers. Very well then, at chess. OK, they can't create images on their own. Well, not very good ones anyway. Yeah, but they can't write a whole novel!

We keep moving the goal. Perhaps we've not noticed we're no longer even in the stadium.

 

[Mad Swede]

It's interesting to me that people are defining human intelligence as being intelligence (undefined) plus emotions.

I find this curious for a couple of reasons. One, outside of these discussions, intelligence is often portrayed as a thing apart from emotion; or, more properly, that intelligence is associated with rationality, and emotions are often viewed as being irrational. If emotions really are our last, best bastion against the machine, I suggest the battle is already lost.

The converse of that is, I don't see anyone arguing humans really are just plain smarter than machines. At least, not in the way the argument went thirty or fifty years ago, when we would brazenly challenge computers to a chess game and crow in triumph at each of our wins. I think as a society (certainly not for every individual), we've already yielded the field.

I'm not persuaded by the logic that points out large models draw upon original human work, without which those models are empty. I can confidently predict that the computers will soon generate far more than we can. Quality? I refer you to Sturgeon's Law, which can readily be extended to computers and humans alike.

Finally, I really do wonder about the distinction between intelligence and emotions. In humans (I'm still willing to say neither exist in computers), both intelligence--reason, rationality--and emotions exist in the brain. I leave aside constructs like spirit and soul for now. Are we really suggesting that these are separate mental processes? There's a clear line to draw? And that one can identify each separately in any given human work?

We readily move the argument to literary work, painting, composition, but that's hardly the whole gamut of human creativity. What about architecture? Is designing yet another warehouse a work of high emotion? How about creating ambient music to be played at the shopping mall, or in an elevator? How much emotion goes into writing a news bulletin?

Oh, computers can't beat us at checkers. Very well then, at chess. OK, they can't create images on their own. Well, not very good ones anyway. Yeah, but they can't write a whole novel!

We keep moving the goal. Perhaps we've not noticed we're no longer even in the stadium.

OK, at the risk of turning this into a discussion of the various psychological theories and concepts about intelligence and self-awareness. Put very simply, human intelligence is usually agreed to be marked by complex cognitive feats and high levels of motivation and self-awareness. It's that self-awareness which is interesting, because that is how an individual experiences and understands their own character, feelings, motives, and desires. And that self-awareness is what lets us authors add the emotional dimension to our work - and it is that self-awareness which AI lacks.