A discussion on Artificial Intelligence and how it's going to affect our industry. 3 3

[DougC]

The current iteration of AI art is pretty lacking unless you spend a significant time refining and rendering singular static images. Current AI has a very difficult time multi-rendering the same background and same character for multi-panel use (from a comic perspective), it is great if you want a single Science Fiction/Fantasy depiction of scenery (without people). I assume this will eventually be worked out if the need arises.

What is going to "save' the industry as it were is that you cannot wholesale copyright AI art. The USCO does allow art modeled with AI, but then you have to publicly declare which parts where made with that AI. Once this happens those parts become unclaimable and essentially public domain. This came to a head back in May with Zarya of the Dawn, they allowed the book to hold a copyright but not the individual images. So I wouldn't get my hopes up of AI Spider-man swinging through the pages until this changes.

Writing though is going to be a very tough gig and will likely transition to an Editor/Clean-up position. The USCO also will not grant a copyright to AI generated text but that is a much more difficult avenue to try and prove. Besides, Stan already shown us that you don't need to write comics; just give the artist a single page (being generous) outline and have them draw something for 20 pages then make up the words later, this was back in the 60's. I can already have AI write me a pretty good outline story that I just have edit and add some additional context and I am sure it is already being used at the highest levels and not being acknowledged.

 

[VintageComics]

On 9/3/2023 at 10:12 AM, alexgross.com said:

AI has already replaced many creative art jobs such as illustratior and graphic designer. how many newspapers now use AI generated images to accompany articles instead of paying someone to make a piece of art? it's sure to replace jobs in the comic art business too unfortunately.

right now, it's still fairly obvious and less than exciting when AI imagery is used in a comic like your friends examples above. it's neat that he can make this all himself, but it does not have much personality to a discerning eye. before too long though, i imagine that midjourney and other apps will truly be able to ape in a non-obvious way the best comic artists working in the field now or in the past. do you like j scott campbell and neal adams? a truly sophisticated AI will be able to synthesize those into a fascinating and possibly even good looking drawing style, hard as it is to fathom. you can even decide if you want a more klaus janson inking look or a josef rubenstein, for example. right now that's not an option, thankfully. but it will happen. 

there's no doubt that as AI continues to improve, it will displace more and more creative jobs in the art biz, movie biz, music biz, comics, etc. it's also replacing jobs in medicine (radiology) law (paralegal) and many other significant high paying fields. 

it appears we will be facing a major crisis when we get to the point where a huge number of jobs have been eliminated due to AI. it's just a matter of time. how we handle it will be one of the great pressing questions of the next decade or two. 

 

As a professional in the biz, I appreciate your insight into the discussion. 

On 9/3/2023 at 11:42 AM, Ken Aldred said:

The art in Roy's example is just as soulless and lazy-looking as the photo referenced art by creators such as Greg Horn was, back in the 90s.  Photo images with a bit of additional rendering on top by the artist.

Still some way to go.

Alex Ross came to mind with his lightboxing. 

--------------------------------------------

To me, this brings on the real question of 'what is a soul'?

And what I mean, is that art is historically meant to "stir the soul" - throughout history, art was an expression of humanity and the soul, and yet some believe the soul doesn't exist.

For the record, I DO believe there is a soul and it is a living thing. 

And I think that the real battle or showdown is already happening between factions who believe there is a soul and those who believe there isn't. 

Pure profit vs pure art. 

Atoms vs order. 

Random structures vs intelligent design. 

In fact, I believe you can draw a string through nearly every faction in human history and will see that it is all divided on this one, single point of whether man is soulless or soulful. 

A 'soulless' person may not even recognize what's missing in art in much the same way that some don't recognize that the soul is missing from a Digital music recording, whereas a 'soulful' person may see the same piece of art the exact opposite way and not be moved by it at all, sensing that it's missing something unique and human. 

I remember listening to a CD for the first time in the late 80s or early 90s and I vividly remember thinking it sounded EXTREMELY different than an analog vinyl disc. 

To this day I still believe that digital is missing the humanity and warmth of analog recordings. 

Edited September 3, 2023 by VintageComics

[Ken Aldred]

On 9/3/2023 at 8:13 PM, VintageComics said:

Can you imagine humans not having the ability to fear?

"Honey, there's a Tyrannosaurus Rex outside out door."

"Oh sweet. Imma go pet it...

Sabretooth tiger cat.

Edited September 3, 2023 by Ken Aldred

[VintageComics]

On 9/3/2023 at 1:16 PM, Ken Aldred said:

Sabretooth tiger cat.

I actually typed that in and then erased it.

I felt like T-rex made my point with a little more emphasis. 

Edited September 3, 2023 by VintageComics

[Beastfeast]

I’ll take your word that your friend is a solid writer but that art/layout/design looks absolutely terrible. 

[shadroch]

Yes, early humans were terrified of running into T-Rex. 

[mrc]

On 9/3/2023 at 8:57 PM, VintageComics said:

I remember listening to a CD for the first time in the late 80s or early 90s and I vividly remember thinking it sounded EXTREMELY different than an analog vinyl disc. 

To this day I still believe that digital is missing the humanity and warmth of analog recordings. 

Generally, I would agree...............

[VintageComics]

On 9/3/2023 at 4:38 PM, Beastfeast said:

I’ll take your word that your friend is a solid writer but that art/layout/design looks absolutely terrible. 

His novels tend to sell well and his 1st graphic novel sold well on Amazon.

We did speak about graphic art and picture story telling. Rendering is one thing. Layout and design is a totally different discussion and I agree that as far as picture story telling, there's always room to improve. 

Not everyone can be a Kirby or an Eisner but most can aspire to be somewhat successful. 

[VintageComics]

On 9/3/2023 at 5:12 PM, mrc said:

Generally, I would agree...............

My heart just skipped a beat. 

Wait, should I be worried? 

[Nick Furious]

On 9/3/2023 at 5:03 PM, VintageComics said:

His novels tend to sell well and his 1st graphic novel sold well on Amazon.

I suspect the graphic novel sold based on his previous reputation and fanbase.  I don't mean to undermine the overall point, but that can barely be called a graphic novel.  More like "pictures added to story".  It reminds me of something I did in 8th grade where we had to write a report and find existing illustrations that went along with the narrative.  In a good graphic novel such as DKR, the artwork is not just illustration, it is unspoken narrative.  It shows, rather than tells, what an aging Batman looks like and how he outfits himself.  It acts as both a second and third dimension to the overall story.  At the moment AI is powerful but generally flat, depending on information that is already available.  There may come a time soon where we get 3D AI, something that will not just recreate any song in the voice of Sinatra but can also recreate the song in the language of Sinatra.  My biggest fear of AI is what happens when we get AI vs. AI?  When one is charged with hacking a website at all costs and another is charged with defending it at all costs?  Who knows what level they will take it to.   

[VintageComics]

On 9/3/2023 at 7:10 PM, Nick Furious said:

My biggest fear of AI is what happens when we get AI vs. AI?  When one is charged with hacking a website at all costs and another is charged with defending it at all costs?  Who knows what level they will take it to.

For the sake of transparency, I came to this question as soon as I understood what AI was. I immediately understood the implications toward mankind. Since the advent of the various public AI offerings, I think people are slooooowwwly coming around to understand it's implications and my reasoning behind starting this thread was that I believe MOST people have never thought about this. 

This isn't "living in fear" as some may like to paint it. It's living in reality.

You have to face the scariest things in life or they overtake you. 

On 9/3/2023 at 7:10 PM, Nick Furious said:

There may come a time soon where we get 3D AI, something that will not just recreate of Sinatra but can also recreate the song in the language of Sinatra. 

A machine can't be human and it will never will be human. I'd bet the house on it. 

But it may be able to be so close to a human that you may not be able to tell them apart. That's one concern. 

The other concern is the backdrop that this happens against.

Humanity is headed in a direction where some / many (most?)  WANT to leave their humanity behind and be something they're not. You can see it with implants and cosmetic surgeries, what people are willing to put into their bodies, the type of music they listen to (I can't listen to fake music - it's soulless), the food they eat (hey, this lab food is better than real food! ).

 

People are not only starting to love fake things en masse but there is a belief on one side that humanity is actually bad and getting rid of it is better. 

A perfect example would be autonomous cars. 

The argument goes that autonomous cars will eliminate so many problems including getting lost, car accidents (so reduction in deaths and injuries), congestion and all of these horrible ills that plague our society. So tech provides what they think is going to be a "good solution" to a known problems. 

But nobody is talking about the OTHER side of the coin. That discussion has been quashed. 

Like, why the heck are people who don't know how to drive allowed to get licenses, for fudge's sakes!

And what will be the societal effect of placing less responsibility on human decision making? Well, the obvious one we see already is a dumbing down of society. As people bring more creature comforts into their lives, it softens them, makes them weaker, makes them poorer decision makers, makes them less intelligent. 

Necessity is the mother of invention and we are devolving as a species as tech advances. 

-------------------------------

Just look at how much unhappiness there is in the world and where it's concentrated. I think you can argue that the most technologically advanced regions also harbor the unhappiest people while the poorest regions may generally contain the most peaceful and content people. 

This just brings me back to my earlier point that this all seems to be split into two schools of thought which believe either that humanity is fundamentally good and is more than just a set of atoms and building blocks but actually has a connection to the rest of existence or that it's just random chaos and you can rearrange those building blocks any way you want and it doesn't matter.

Yet we have historical records that have already dealt with this dichotomy and every time society is dealt with like a bunch of lego bricks, it crumbles and fails. This a pattern that has perpetually repeated itself countless times.  

 

What does this have to do with this discussion?

AI art is just a small piece of a large machine and how it gets accepted by the public should be determined by the audience but it will likely be pushed in spite of the audience's wishes, because of the implications of those who control the AI:

In short it will be more profitable for companies to use AI than not, humanity be damned.

And so I believe those companies will try to culture society to accept it even if they don't.  

 

Edited September 4, 2023 by VintageComics

[Ken Aldred]

Emulating comic art at a convincing level seems quite complex.

The basic linework, figure proportions, facial expressions of an artist, say, of Kirby’s, maybe, but then there’s the artist’s use of perspective, figure placement, camera angles, the page’s storytelling and pacing layout, when to use more dynamic effects or to rein it in and create more grounded images, and characteristic background and other environmental details.

A lot for an algorithm to account for to avoid an unconvincing and Uncanny result.

Edited September 4, 2023 by Ken Aldred

[valiantman]

We'll see the average replaced first.  People who do an average job should always worry because there are always replacements for people who are average, whether it be other average people or technology. Furthermore, people who do an average job have always been at risk of being replaced by someone or something which does a better job.

I am not convinced that people who do a job well have anything to worry about yet, whether it's creating universes in fiction/art or driving a truck.

Edited September 4, 2023 by valiantman

[Nick Furious]

On 9/3/2023 at 11:23 PM, valiantman said:

We'll see the average replaced first.

I was kind of thinking the same thing.  The filler, the stuff that doesn't capture our imagination or take the mediums to new places will be pretty easy to replace.  

[Aman619]

This argument is an alarmists joke. Just ask the attorney who filed a brief he had written by ChatGPT. It used precedents cadged off the internet, like Moby v D ick…   All of them were caught by the court and he faces disbarment.

the problem with these AI tools - large language programs - is that they are trained by assimilating everything on the internet.  Which means they are learning from garbage and falsehoods along with true reporting… but cannot tell the difference! They lack the “common sense” people learn by the time that are five years old. 

And AI created content is already a significant % of the internet content and has been become a part of the data that AI is still “learning from” and devising answers with. Further diluting it’s ability to give correct answers 

It gotten so bad already that the AI software companies have preserved a state of the internet from BEFORE IT WAS POLLUTED BY AI created data. So they have a “purer” base of “knowledge” to work with for futures iterations.  
 

get your pitchforks if you must. And loudly proclaim the end is nigh if that makes you feel special or some kind of genius who can “see the future”. While the rest of us just accept it as a new tool to use and keep our eyes open as to its pluses and minuses. 

Edited September 4, 2023 by Aman619

[valiantman]

On 9/4/2023 at 2:25 AM, Aman619 said:

The problem with these AI tools - large language programs - is that they are trained by assimilating everything on the internet.  Which means they are learning from garbage and falsehoods along with true reporting… but cannot tell the difference! They lack the “common sense” people learn by the time that are five years old. 

It's pretty funny when you think about it.

Imagine running everything on the internet about "bigfoot" through a computer that is doing calculations about how often words and phrases are used on the internet or whatever sources it is compiling.

The computer will note that words like "monster", "yeti", and "sasquatch" are commonly found on pages that include the word "bigfoot", maybe even a phrase like "abominable snowman"... so it associates these words and phrases with "bigfoot".

Another phrase it will associate with "bigfoot" will be "monster truck", and since "monster" is also a common description of "bigfoot", it makes sense that "monster truck" will be associated with "bigfoot". 

The concept of "large tracks" works for both "bigfoot the monster" and "bigfoot the monster truck".  Both may leave "exciting evidence" of physical destruction in an environment, rather than direct evidence.

At this point, the AI (in simplest description) might be ready to declare that "bigfoot" is also known as a "yeti", "sasquatch", "abominable snowman", and "monster truck" which leaves large tracks through "the Pacific Northwest area of the Himalayan mountains in Canada and Scandinavia". (Because these geographies are commonly mentioned as well, they are as-far-as-a-computer-can-tell likely to be a general description of the same area.)

There you go, rinse and repeat... that's AI.

It takes a completely different type of program to take these AI "bigfoot" statements and break them apart for AI fact-checking by AI ("truck" doesn't belong, the Himalayan mountains aren't in Canada or Scandinavia, the Pacific Northwest usually includes Washington, Oregon, non-Canada areas, the tracks are completely different but both are large), but these combined statements made by AI have never been problems discussed on the internet before AI. Where is the reference data to tell you that large tracks from trucks aren't found on mountains on the wrong continents as evidence of a theoretical but unproven monster?  That's not a "learning data set" that's available.

You would need AI to tell you about trucks, AI to tell you about mountains, AI to tell you about countries and geographic regions, different AI to separate the cryptozoology and automotive term "tracks", and each of those AI training databases may have its own problems similar to how "monster truck" ended up in the "bigfoot" AI results in the first place.  At least four error-laden AI systems would need to be used to accurately fact-check a single "bigfoot" topic's AI results without creating any new errors that would need additional systems to correct, all while receiving many sources declaring that bigfoot both does and does not exist.

(TL;DR - AI is a Sim that can't get out of the pool)

Edited September 4, 2023 by valiantman

[namisgr]

Machine learning as a part of AI is more sophisticated than that, but requires a lot of specialized and tailored development, and isn't likely to enter the comic book field any time soon.

Machine learning is an umbrella term for solving problems for which development of algorithms by human programmers would be cost-prohibitive, and instead the problems are solved by helping machines 'discover' their 'own' algorithms, without needing to be explicitly told what to do by any human-developed algorithms.

 

Edited September 4, 2023 by namisgr

[VintageComics]

On 9/4/2023 at 12:23 AM, valiantman said:

We'll see the average replaced first.  People who do an average job should always worry because there are always replacements for people who are average, whether it be other average people or technology. Furthermore, people who do an average job have always been at risk of being replaced by someone or something which does a better job.

I am not convinced that people who do a job well have anything to worry about yet, whether it's creating universes in fiction/art or driving a truck.

This is a great point. 

That lawyer above is about to get replaced. 

----------------------------------

But calling AI archaic and "nothing to worry about" at THIS point in time is probably accurate. 

But that's like looking at an early cell phone and Ma Bell saying  "Brah, you got NOTHING to worry about. This thing doesn't fit into your pocket, it's expensive, it doesn't work.... It'll never catch on." meanwhile now we're talking about humanoid implanted cell phones.

Hmmm......

I REALLY think many can't see where things are going before they actually get there and the apple cart falls over. In fact, after the last few years of heated debate, I'm damn 100% of it now.

Edited September 5, 2023 by VintageComics

[VintageComics]

On 9/4/2023 at 3:25 AM, Aman619 said:

This argument is an alarmists joke. Just ask the attorney who filed a brief he had written by ChatGPT. It used precedents cadged off the internet, like Moby v D ick…   All of them were caught by the court and he faces disbarment.

the problem with these AI tools - large language programs - is that they are trained by assimilating everything on the internet.  Which means they are learning from garbage and falsehoods along with true reporting… but cannot tell the difference! They lack the “common sense” people learn by the time that are five years old. 

And AI created content is already a significant % of the internet content and has been become a part of the data that AI is still “learning from” and devising answers with. Further diluting it’s ability to give correct answers 

It gotten so bad already that the AI software companies have preserved a state of the internet from BEFORE IT WAS POLLUTED BY AI created data. So they have a “purer” base of “knowledge” to work with for futures iterations.  
 

get your pitchforks if you must. And loudly proclaim the end is nigh if that makes you feel special or some kind of genius who can “see the future”. While the rest of us just accept it as a new tool to use and keep our eyes open as to its pluses and minuses. 

You're standing on the side of the freeway and seeing cars moving and assuming there isn't an accident and based on that, there never will be. 

Is AI learning more or less each day?

Is it getting faster or slower?

Is it's base of influence getting larger or smaller?

There are no pitchforks. I just believe the conclusion is inevitable but this is just a discussion...and I happen to have a pretty good track record on being right in discussion. 

 

[MatterEaterLad]

On 9/4/2023 at 8:51 PM, VintageComics said:

Is AI learning more or less each day?

Is it getting faster or slower?

Is it's base of influence getting larger or smaller?

I was at the Aspen Ideas Festival a few months ago and AI was the unofficial topic of the entire week. 

There were talks by people like Eric Schmidt (Google), Daniel Huttenlocher (MIT), and the list goes on and on. The consensus was that we will have some form or general AI in FIVE YEARS. This blew me away since earlier predictions were that we were maybe 10-30 years away. 

There were debates about whether a general AI would/should have First Amendment rights.