Transcription: AI & Art: Creativity + Law – A Conversation
Transcription
Anne Kelly:
Artificial intelligence, is it a fad or is it the future in terms of art making? There’s a lot of opinions out there. I’ve asked two professional artists and a lawyer to share some of their insights. Are you curious what they had to say? We’ll, keep watching.
Ed Bateman:
Hi, I am Ed Bateman. I’m an artist and a professor. Anything having to do with art totally intrigues me and fascinates me. I can never get enough. I’m
Talia Kosh :
Talia Kosh and I’m an attorney in Santa Fe. I work with artists and filmmakers and intellectual property in general.
Ruben:
I am Ruben. I’m a photographer. I had sort of a long history in creative technology, places like Lucas Film and Adobe, and now I’m in Santa Fe teaching photography. I’m a collector and like anything to do with photography and technology in
Anne Kelly:
Particularly. I’m so excited to have everybody here think it would be fun. Since we’re talking about artificial intelligence in art, if we as a group opened up one of those texts to image generators and created something together, what do you guys think? Sound fun? Yeah,
Speaker 2:
Sure. Yeah. Our attorneys here because I’m not really sure if the four of us make it in Dolly or something. I know where the rites go.
Anne Kelly:
I haven’t actually used this one before. Night Cafe. Anybody else? I have not. I’ve only used Dolly.
Speaker 2:
I’ve even heard of Night Cafe.
Anne Kelly:
That’s kind of why I thought we’d try it. I’ve only used Dolly myself at this point. Prompting seems to be a skill.
Speaker 2:
I’d go farther and say the skill is the prompt. That’s the artwork. You can go in and write, show me a car with a tree growing out and make it a 57 Chevy. But very quickly you start realizing the nuances in the order of the words, the kinds of words you use. And I think like a lot of technology, there’ll be people who get good at creating prompts for things for people.
Anne Kelly:
People are selling prompts already see
Speaker 2:
People sharing prompts. I see a lot of community work where people posted sheets of really interesting, complicated from what position it’s answering the question. You are a fourth grader and you only speak in Latvian From that context, what I want you to do. And then it makes it different than if you’re an ARC professor. And I found it’s not particularly good with metaphor and visual language as an experiment, I entered in the first stanza of the Battle Hymn of the Republic because I thought it had very evocative language, trampled out the vintage and seen the glory of all of that stuff, and it was dreadful. Anything that came about, so metaphor, but again, we’re at the very earliest stage. You may have been using chat GPT-3 or an earlier version and already version four, which came out yesterday has so much more nuance in what it can understand.
Speaker 2:
You can feel how these are ratcheting forward very, very quickly in the sophistication of the language you can use. It was super literal in version two. It just did what you said and it couldn’t think out of the box. And now it’s really pulling a lot in. And I also think with the battle him of the Republic, I think it blocked some of the images because it was thinking that there might be some gruesome imagery or something that’s inappropriate for it to render. I gave Midjourney Westin’s pepper 30, I fed pepper 30 in and wanted it to use something about the nuances, but it perceived that as pornographic and it would not take it. The bell pepper was too sexy.
Anne Kelly:
Oh, that’s funny.
Speaker 2:
I love that. Yeah, I thought that was interesting.
Anne Kelly:
So one of the things I think is interesting from the origin of AI is this concern, well, kind of a truthful thing, this concern that artificial intelligence will become more intelligent than humans, and that’s a scary thing, but one of the challenges it’s had over the years is even if it’s been able to read everything on Google, for example, it doesn’t have the benefit of human experiences. I tried to create an image earlier, and I guess it wasn’t specific enough. I requested an image of dogs eating tacos and smoking, and it was clear to me they’re smoking cigarettes, but the dogs were actually kind of on fire that they were eating tacos, which was really kind of fun. But it is an example of how there’s certain things that it doesn’t understand yet, and I guess that’s where the prompting is helpful. So I tried it again with smoking cigars, very literal, right? There’s certain things that it doesn’t depict very well, like diamonds. There’s certain objects that it’s not capturing very well, including faces.
Speaker 2:
Hands, also.
Anne Kelly:
Hands, yeah. They’re like weird clone hands,
Speaker 2:
Lots of weird fingers. They just put fingers all over them.
Anne Kelly:
And the styles are important, right? Certain styles generate better imagery than other styles. I feel like we should use some of those. You’ve mentioned diamonds and hands and all the things that it has a hard time with. How about a
Speaker 2:
Handholding? A handful of diamonds and jelly beans and I think we should say a photograph. Yes. Photorealistic image. Maybe a woman’s hand. How specific? A well manicured a wealthy woman’s hand. You go in down that rat hole, right? Holding a giant diamond Smoking. No.
Anne Kelly:
Does this prefer that you put the style at the beginning or the end? Does it matter? I’m not sure. I haven’t used this one before. I only have a few credits and then you have to pay, but we’ll see.
Speaker 2:
I could take us into Dali if we run out of credits here. I think I’ve got some old prompts,
Anne Kelly:
Diamonds and
Speaker 2:
Something weird. Something that you wouldn’t put with diamonds.
Anne Kelly:
Diamonds and cats in
Speaker 2:
Your hand or ants. I was like ants. Diamonds covered with ants. Yeah. What’s opposite of diamonds? Yeah, charcoal. Cole,
Anne Kelly:
Let’s go with ants.
Speaker 2:
Okay,
Anne Kelly:
Anything else? That’s arts. That’s
Speaker 2:
Arts.
Anne Kelly:
Cool though we could see what their interpretation of arts is. We
Speaker 2:
Probably have to put apostrophe S on the woman. So it’s not
Anne Kelly:
Women’s. You’re right.
Speaker 2:
Is it a close shot? Is it wide? Is it, just try it. How about say ants and a green pepper? Talia of course where I go to, even as I said I was giving pepper 30 to the computer for it to copy the style. What laws are being infringed upon there?
Anne Kelly:
What issues does that, Ricky? Let’s make it’s thinking about it. I
Speaker 2:
Should hope so,
Anne Kelly:
Because right now it’s going through its database of images, not bad. Second glance, frightening.
Speaker 2:
You can see the fingers are a little bit,
Anne Kelly:
They’re all warped and distorted. Okay. There’s the first one. It makes a few, it’s like a diamond bracelet. The pepper kind of took her predominance.
Speaker 2:
I like the ant off to the left. The ant is photo bombing.
Anne Kelly:
That’s pretty good. And then we have, oh,
Speaker 2:
Wow. Well, for the moment our careers are safe, but
Anne Kelly:
I think I can sell this to anybody.
Speaker 2:
So here’s the thing, Anne, you can’t sell it to anybody, but what you can do is if you are writing an essay or a blog post on ants and diamonds, you might find this delightfully weird and fun and it satisfies what you need an illustration for. In your blog post, you wouldn’t have hired someone. You’re not capable of doing it yourself. You didn’t want to go get a stock image. You’d never find one of diamonds and ants. And this is plenty evocative. Provocative. And so that’s how market has been increased. It’s not like you took this job from somebody, you never would’ve hired them, but you’ve now done that kind of job. That’s what I would say. I
Anne Kelly:
Love how AI takes its licenses because you didn’t say anything about a Mel or
Speaker 2:
Bad
Anne Kelly:
Teeth or an emerald. It turned into an emerald when you mixed a green pepper with a diamond. Yeah. Or the teeth. Yeah. It’s just very interesting.
Speaker 2:
It looks like caviar in her mouth sort of freaking me out a little bit. Again, not specified and
Anne Kelly:
Those nails,
Speaker 2:
But if you saw this in a blog post, it might stop the clicking. It might pause you. It might do what people want images to do in their posts, in their articles. Sometimes that’s weird. Part foot, part hand. Are those raisins on the
Anne Kelly:
Bottom? That’s a pile of am.
Speaker 2:
Oh,
Anne Kelly:
I am just guessing. I wish it would tell you what they think it is. What this computer program thinks it is.
Speaker 2:
Is there a candy? Is it called ants or something that looks like that? Or is there some little snack food Anyway?
Anne Kelly:
I have no idea. Well, this could end up being the whole episode, but do we want to try one more real quick and then knock it out?
Speaker 2:
Sure. I kind of do. Or do we want to edit this one? Yeah, so usually you just keep tweaking your prompt as it iterates closer to what you’re going for. Let’s
Anne Kelly:
Do it in the style of Van Gogh.
Speaker 2:
Let’s do Edward Weston. We’re photo people. We are photo people. I
Anne Kelly:
Find that when you’re not doing photo realistic, it comes out better.
Speaker 2:
Yeah, I agree.
Anne Kelly:
Let’s just change photo realistic to
Speaker 2:
Painting by Winslow Homer Van Gogh of, yeah. I tried to create a new Cindy Sherman work, a new early work of hers, and it was funny. It was drawing upon older images of her.
Anne Kelly:
Do you think it Nas Man Ray is, it
Speaker 2:
Knows a lot. I don’t know what this knows, but Chachi knows who man Ray is. Man Ray. Although I bet you it confuses Man Ray with some kind of man in a Reagan or a photograph in the style of Man Ray, you learn the language that they kind of like
Anne Kelly:
Anybody want to add anything?
Speaker 2:
A woman or a woman’s hand? Singular.
Anne Kelly:
Oh, right.
Speaker 2:
I never know how much, I mean, that’s what community rooms are. People sharing tips about language tweaks that make things happen better. Like oh, put the style at the end. Oh, put it at the beginning. Oh, because it skips some of the stuff at the end. Sometimes it doesn’t get all the way down the pipe or whatever. I’ve
Anne Kelly:
Found a few times I’ve given it too much information and it just skips some of the things I don’t know. Well, let’s see if it knows who Man Ray is.
Speaker 2:
What’s that?
Anne Kelly:
They’re curious
Speaker 2:
Fingers, man.
Anne Kelly:
Does anybody know that Man? Ray photograph? I
Speaker 2:
Think I missed that in the history of Man Ray, but we’ll make it black and white. It’ll look better. Yeah, it’ll look more man. Ray ish. What have we got
Anne Kelly:
There? Look like little cartoon characters. The ants as they look like
Speaker 2:
Sea monkeys. And
Anne Kelly:
What are a green manicure?
Speaker 2:
Yeah, or the French tip? The green French tip tip.
Anne Kelly:
Oh, that’s some fabulous fashion photography.
Speaker 2:
That’s love. Yeah, heart. Interracial intergalactic where diamonds mean different things to
Anne Kelly:
And
Speaker 2:
Green pepper fingers.
Anne Kelly:
Oh, you were right about that inner planetary relationship.
Speaker 2:
And how did it get that from the prompts? If you read the prompts, there’s no way what the,
Anne Kelly:
Yeah, I have no idea. There’s no diamond in this one. Good point. Preferred the pepper.
Speaker 2:
Oh, y’all.
Anne Kelly:
Yeah.
Speaker 2:
It’s almost like it thinks that man Ray is an alien man. Yeah,
Anne Kelly:
That’s how it interpreted
Speaker 2:
It. Oh, Ray. Like Ray got it. Really did. Yeah, yeah,
Anne Kelly:
Yeah. Right.
Speaker 2:
Or we could do a man Ray picture of a man Ray.
Anne Kelly:
All right, well, anybody else want to tweak it or shall we hop out?
Speaker 2:
I’m typing this into Dolly to see what it does turn the style of a man, ray of a woman’s hand, diamonds and ants and a green pepper. Let’s see what that does. Switch the screens here if it’s worthy. Oh, better here. Want me to share this?
Anne Kelly:
I think that’s definitely closer to man. Ray Dali doesn’t understand what a diamond is.
Speaker 2:
It doesn’t understand what a diamond is you’re writing. Interesting. What
Anne Kelly:
These programs are doing is scraping the internet for imagery in order for it to learn the dataset. But that doesn’t mean it’s processing and understanding. Right. There’s a different,
Speaker 2:
This is the modal Lisa in crayons by a 6-year-old, not bad. Here’s origami in a white sunlit room. It occasionally surprises you.
Anne Kelly:
And Ruben, I remember you telling me about this program last summer. Was that before it was released or was it just really early?
Speaker 2:
It was a month or so before it came out. They were just starting to open it up a little bit more. I just wanted to see what it would do. I mean, I’m always curious of things that I think are going to destroy photography or improve it ed, whatever. But I’ve always felt it’s been a slippery slope since media got digital by being really negative. I’d say Photoshop was the first nail in the coffin and then this pushed it off the cliff, but it’s just changed picture taking. I think for commercial photographers whose job it is to make a product look good, to convey something for a client where they have objectives, you’re still at using a camera is almost the smallest part, if not even not a part. You may need to get a photo of some part of this as a background plate or an element, but ultimately you’re going into Photoshop with or without AI and you’re putting things together and you’re making that product look great and you’re putting the sun coming in over it and all of that is art.
Speaker 2:
I’m not even going to argue that it’s not art. All of it is creative and fun and makes for really cool product shots. If they’re done well, they’re indistinguishable from the work a photographer might do, but I can do it a lot cheaper. I can do it a lot faster as a client. I get a lot more control and the job of the photographer. It reminds me when you’re doing special effects in movies, and you can probably speak to this too, where the effects coordinator looks at the script and says, we need a bunch of dinosaurs eating a car. Let’s look at the budget. Can we do it with 3D models? Can we do it with a mechanical dinosaur with a back plate? Can we do it with miniatures? What’s going to give us the right effect for our budget that fits with this stuff?
Speaker 2:
And it’s really not photography. I mean you might call it cinematography there, but it’s a lot of stuff. But I think what we think of as photography isn’t, that is a different job if you want to go off and be a commercial photographer, if you aren’t a master of these tools, Photoshop, AI things, you’re not going to work. The fact that you’re very skilled at using your hassle wa and capturing the light is irrelevant at going forward. That’s what I would suggest. Or usually it’s parts of teams anymore. The photographer has their retouch or person, their stylist, sometimes their lighting person. So it’s already that way in some respects, which is why I think it’s not photography, it’s composition, it’s art. That photo elements may or may not be a part of the final photographic product. It could be a team or it could be an individual, but it’s not what I think of as photography.
Speaker 2:
I’m not talking about commercial. Commercial stuff is going to be all computerized. It’s only art photography that is going to be purposely constrained using an old fashioned camera with film eight by 10 camera, an ethic that says I’m not going to retouch it or add or subtract anything that’s got to come from journalistic ethics and it’s no longer kind of constrained by technologies. Well, and talk about the fine art part. Most of these programs really are focusing on images where the subject is the main element and the arts, that’s often the least important part of an image is what the picture is of. Think of, oh, a Cartier basant image. It’s not what the pictures of that makes it interesting. It’s the composition. I’d say it’s the moment capture. Yeah. Okay. Yeah. But it’s the moment captured made maps of these compositions what he was trying to do, not just a moment but a moment where everything lines up.
Speaker 2:
I know, but it wouldn’t be photography, certainly wouldn’t be Car T broan if things didn’t line up. And so he just went into Photoshop and moved it over and put a bird right there and put that person right there. That belies the entire thing he’s doing. It’s not just the composition, it’s that he caught the confluence of these random moments and found coincidence. And that is part of what I would call photography. And the minute you take it out of coincidence and you just create coincidence. It’s another thing and we’ll disagree on this, but our comments on the cardio basant are total in alignment. And what’s interesting is if your AI machines are essentially looking at precedent images, I don’t think they’re going to be able to pick that up because that kind of alignment is so situational and it’s also so really complicated to parse.
Speaker 2:
And again, like I said, I’ve made little charts of how the eye moves through these images and so you can really see the brilliance of the mind at work in that. I don’t think an ai, at least the way it’s configured now, looking at other images is going to be able to do that kind of compositional sophistication. But I think you’re underselling it. I think we’re only on day one and in six months or two years it may totally understand what the idea of coincidence and in fact the randomness of what it creates and as the driver of the system still look at all the things that produced and said, oh, it may not have meant to do that, but look at this one. That’s kind of a funny little weird thing. I couldn’t have done that on purpose and it doesn’t know it did it, but there it is. It’s cool. And I love that one. I think that’s a different issue. The randomness of elements very different than composition composition’s a pretty sophisticated idea.
Anne Kelly:
I think it’s helpful to frame this issue kind of an legal perspective since we’re talking about what modicum of creativity does it take to have a creative work and a piece of original work. Since you’re talking about photography in the late 18 hundreds, the Borough Giles Lithography versus Cerone case back then they were questioning photography like we’re questioning AI now. And the question was, well, anyone can take a picture of any subject. And so how is there enough creativity to afford a copyright to it? And again, they went to what you guys have mentioned, which is there is sufficient creativity. The photographer is choosing the angle, the lighting, the shade, the timing, the choice of subjects, the objects. It’s all in the framing. So moving that into ai, the question becomes is AI copyrightable? And currently the copyright office says, no, only human authorship is afforded a copyright.
Speaker 2:
Is this why you brought up the monkey case
Anne Kelly:
And the monkey case? Yes. Rudo v Slater PETA tried to sue on behalf of a monkey, told the camera from the photographer and took a photo and PETA was trying to claim with the copyright office that the monkey was the author and not the cameraman. And of course, both the court and the copyright office only acknowledge human authorship. So the question then becomes, what if you have a mix of AI in humans? One of the other lawsuits that’s out right now would’ve been created an entire comic book with AI and then registered it in the copyright office but didn’t disclose that it was completely AI made and they initially registered it but then deregistered it once they learned that it was only AI generated. But basically the copyright office says if it’s applying the same principles, if you’re selecting or arranging in a sufficiently creative way, so you’re taking those AI elements and you’re actually selecting and arranging them, you can have copyright on the human authored aspect. So
Speaker 2:
Music sampling has evolved over the past couple decades so that if you take a piece of my song or you borrow a riff from it, the way attributions and copyrights are handled has been worked out because you generally know who you got it from, where they recognize their thing. But in the ai, it’s this, there’s no person there that it’s borrowed from lots of people even if you recognize your work in
Anne Kelly:
There. Yeah. The newer lawsuit that comprises comic book related illustrators writers, they’ve brought an action against, let me see, I can’t recall which one it was. Let me just pull this up. A stability, AI midjourney and deviant art. And so basically they’re alleging that these companies have committed copyright infringement in training their generators, but with millions of images. And since copyright infringement is determined on a case by case basis, this is a really hard lawsuit to win win because in order to prove infringement, it has to be substantially similar. You can’t infringe on a style, you can’t infringe an idea. Those are freely exchangeable. It has to be an expression of an idea. Getty has a lawsuit right now about thousands of their images being infringed upon by these AI generators. They might have a better case because of the quantity of images that they have versus smaller creators who are going to have difficulty proving that their small amount of images or even maybe they’re just dealing with one image that they’re claiming is infringed upon that the AI generator took and created a derivative work of,
Speaker 2:
So in my case where I uploaded or tried to upload Westin’s Pepper 30, that’s a clear case where I gave it this and said riff off of this, but I’ve done similar things where I just said, in the style of Ed Western,
Anne Kelly:
It has to be a specific work because you look at the four corners of the work and you ask on a case by case basis, is there substantial similarity, all of the different elements that go into that. And that’s where it becomes difficult even if you get to that answer and say, okay, yeah, we see substantial similarity. We see that you’ve infringed and if you want to pull up that Getty photo, that’s a really good example because Getty is suing right now. Also stability, ai, which I have not tried. It must be a really good one. It’s also bringing up the watermark of Getty. Look at that. When we look at these two in comparison, the question becomes are they substantially similar so that the copy could be a market transplant for the other? So are they taking the market away from the other one, but
Speaker 2:
They’re not. It looks to be like parody and that’s quite protected. Right,
Anne Kelly:
Exactly. And then the next analysis becomes as at fair use. Right. So most generally does the purpose and character of the use change, right. So we’re looking at this Getty image. The irony is that Getty doesn’t own a ton of its images that it claims it has license over. It doesn’t actually have a license over this particular one. It does, but there are many artists that it claims it has a license over and it doesn’t. There are several lawsuits and claims against Getty for this very thing. So they’re accusing the AI of doing exactly what they do, but they want to be the one that is exploiting as many images as possible in sending out these cease and desist by our license. Even if we don’t actually have the right to license this image. Most people don’t even ask and they’ll just pay the license fee
Speaker 2:
If they don’t defend the pictures, they further lose some authority in this. They’re letting people rip off the pictures that they say that they own. So they
Anne Kelly:
Could make that claim at any time. But yeah, I mean you’ll have a statute of limitations issue at some point, but you could always say it’s an ongoing infringement and go after these folks. But again, they’re also making a claim for trademark infringement. And I would say nobody would look at the second image and say they’re creating confusion in the marketplace. This is the kind of analysis that everyone needs to do. And again, fair use is going to be a very interesting case by case basis of has the purpose transformed. So the purpose of the initial image, I assume was journalistic purpose photographing this soccer game for public consumption. Getty didn’t take the image, Getty just licenses. The question becomes what is the purpose of the AI generated image? And it could be so transformed or so different that it’s considered fair use and that’s what these cases will be arguing moving forward and it’s going to be very case by case,
Speaker 2:
The whole appropriation route. So even the transformation could be pretty minimal. Simply go the Richard Prince route. Just add your own caption to it.
Anne Kelly:
Yeah. What Ed is talking about is the Richard Prince case where Richard Prince took images off of Instagram and commented on them and then took a photograph or a screenshot of those and then blew them up really large and put them in a gallery. And that was based on his initial lawsuit that he won in the second circuit where they said, you don’t have to comment on the original work. It can be a social commentary, it can be anything that transforms the work. And truly, if you look at that, let’s say somebody’s posting a selfie on Instagram, what is the purpose of that post? It’s personal purpose. You’re posting a selfie, but if Richard Prince is commenting on that post and some strange bird talk that he does, making it larger three or six feet and putting it in a fine art gallery, even though it is unfair, the law is an ass and he could probably get away with it with fair
Speaker 2:
Use. Wow. I’m looking at the Richard Prince thing. Wow. I haven’t posted many AI images on Instagram, but before I did that I ran through Google image search just because I was also curious to see if it would find any of them. None of them turned out, but things have been obscured enough. The machine’s not going to notice that. It’s sort of like the kids using chat GPT to write their papers and then the teacher’s using chat GPG to check to see if written papers have been plagiarized or they can tell if the computer has done it at some level. When I was at Adobe, I always felt that at some level it was Adobe’s responsibility to invent the antidotes poison that Photoshop creates, which makes it impossible to know if an image has been manipulated. But I worked with a lot of those guys. It’s impossible. Ultimately you could just take a picture off your screen with a camera and then you lose all metadata. So it’s just not trackable. I think I had a paper for my history photography class last semester. I didn’t have the time to go too deep, but I suspected was generated that way partially because the large first paragraph was referring to Julia, Margaret Cameron by her maiden name.
Speaker 2:
I could identify other little snippets of the text from where they were drawn from. That doesn’t surprise me too much, but the Julia Margaret Cameron’s maiden name was rather a surprise.
Anne Kelly:
So when you were grading that paper, did you just write chat GPT on it?
Speaker 2:
Actually, no. I’ve phrased this question in a way that’s very hard to simulate. Essentially it’s an imaginary conversation between Julie, Margaret Cameron, Henry Peach Robinson and Carlton Watkins. They’ve all gathered, they’re all having dinner. It’s a fun one to write I think, but it’s a hard one to sort of simulate. I think for a machine, at least at this point, they all have different styles. They debate all the pros and cons of their life and choices and work. That’s a cool question. I want to go to college again. That sounds
Anne Kelly:
So there’s another question for you, ed, as a professor, as someone receiving papers that students may have written with chat GPT, do you mind, as long as they’re using it as a tool, a collaborator as opposed to say just having it generate the entire paper
Speaker 2:
If it was interesting. It’s just that these things are organizing information, but I don’t think they’re showing any discoveries, making connections and links between things that I’m aware of. Do you really want to spend your time reading papers that students didn’t spend time writing? Doesn’t that feel weird? I’m going to sit around and read what the computer wrote and give it a grade based on how well you discerned whether the thing it wrote was good. We’ll turn the thing around. What about teachers having AI software? Would I trust software to do the grading on these kinds of papers? I would love that. I mean, 44 papers, 300 words is a lot. But I also know that I can’t make a simple rubric to share. It becomes a plugin question, but would I trust a machine to grade those? Maybe the level of effort involved, but I kind of don’t trust it and to make meaningful comments on what they’ve written. I think
Anne Kelly:
In a law school exam it might work because the graders are just looking for keywords, right? Yeah. So you’ve said all the right words on this, and that’s five checks and that’s five out of five for that question. So I think depending on the subject matter, it might work. And GPT just passed the Google coder exam. Really fun colors apparently.
Speaker 2:
I’m sorry, where my mind goes. There’s a scene in some movie, I don’t think it’s Ferris Bueller at some other movie, but it starts with the professor talking and his student has left a tape recorder on his desk. He can’t be there. And then it’s a week later and now there’s a lot of tape recorders in the audience, and now it’s a week later and the whole room is tape recorders. And now the last saw is the teachers a tape recorder playing and their tape recorders are recording. Beautiful classroom. Something about that is this.
Anne Kelly:
Yeah, that’s what it feels like.
Speaker 2:
But now I’m intrigued. The grading would be nice. Just the images though. I mean, can you ever imagine, and you do a lot of portfolio reviews where a machine, we’ll just call it a machine, could maybe do the job as well as you or others. Well,
Anne Kelly:
You could turn me into a chat bot if you’d followed me to enough portfolio reviews and saw the work that I was looking at and the responses I gave and may teach it, all of that. And then you could just do a portfolio review with my chat bot
Speaker 2:
In the past week, my friends have taken one episode of my podcast, fed it in it, now mapped my voice, intonations and language, and then they are typing things and it’s reading them like me. And that’s today my children. We have a secret code between us because even looking at me on Zoom and hearing my voice, not really the sure thing. So it’s changing very fast.
Anne Kelly:
It brings to light the ethical nature of this, right? I know a lot of people hate him, but he is accurate in terms of ai. Elon Musk said it’s the biggest risk to humankind. And even though there’s growth of global AI adoption, there’s no federal legislation on AI in the United States. It’s just kind of this patchwork of various frameworks. There’s a few code of ethics you can adopt it or not, but we need to be talking about this in terms of privacy regulation, make sure that it’s trained using data sets that are representative and not misrepresented of information. There needs to be a mandatory testing before deployment because if you have AI that’s going awry even in the collection of people’s data. One of the lawsuits now is an app that collected people’s facial recognitions to create an avatar, and they didn’t know that it was doing all of this biometric surveillance of them. So we’re getting into really concerning territory when we don’t have laws around this stuff. In terms of accountability and governance mechanism, I
Speaker 2:
Think the genie’s out of the bottle and it would be great to have some regulation here, but bad agents can do what they want. And unfortunately, I think the problem is that we connected every human with a digital pipeline between us. We made it so everyone has a voice and everyone can speak, and that is how we get all of our information. And it seemed like an amazing thing to put every type of media through this pipe and now you can change everything coming through the pipe. So you never know if what you’re getting through the pipe is real or not. And consequently, it’s like you poisoned the well. And I don’t think regulation will stop it. We are all connected. It only takes one tweet from one psycho to get a whole bunch of people panicking about something or another. And I’ll add a phrase from one of my friends because we’ve all seen the movies, robots, they just go bad on you.
Anne Kelly:
I’ve been randomly recording people on the chairlift asking them what they think about the current state of artificial intelligence. And I think every single person has mentioned the Terminator
Speaker 3:
Became self-aware as Skynet in August of last year. Wait, is that the Terminator movie?
Speaker 4:
And the different things that we’ve seen have only been the tip of the iceberg of what
Anne Kelly:
Could have things like with artificial intelligence. People either think it’s going to save the world or it’s going to destroy the world.
Speaker 2:
It’s the AI that will do it. I don’t think it’s Skynet that comes and kills us. I just think it’s someone else using it as a weapon. That’s my thought. As opposed to it getting sentient and deciding it doesn’t need people. I’m not as worried about that as I am. North Korea infiltrating systems and wrecking havoc on systems.
Anne Kelly:
He has created some really disturbing answers about what it wants to be doing and what it thinks about the human race. So I don’t know, trying to find it for you, Ruben, but it doesn’t look favorably on the human race and it doesn’t want to be doing what it’s doing.
Speaker 2:
I have this little thought in my head because in some ways, if you think about it, we’re at the stage now where it’s really like the infinite monkeys of infinite typewriters working away. We are the ones still making the choices, almost still a curatorial process in a way. Well, we’ll keep generating images till there’s one that I like. The one thing I kind of imagine is that somewhere in some part of the world there’s a bunch of exploited laborers sitting at Photoshop rooms with thousands of these people getting the prompt hurrying up and getting it and sending it out. Sort of imagine. I mean, it creates, we’re still having to search through all of the images to find the hamlets that have been tied at this point. I go down this sort of negative pipe sometimes. I think Bill Joy, one of the founders’ son, got very negative about the internet because, and this just pushes it even farther, but it’s like it used to be start a war or to kill a bunch of people.
Speaker 2:
It took a lot of people. You had to rally people to a cause and it would take a lot to do a lot of damage. Slowly it’s become asymmetrical where one individual, five individuals out of the billions on earth can disproportionately destroy every system. You don’t need a lot. You don’t need a political movement. You just need a couple crazy people at the end of the block who were bored enough and technically savvy enough. And that’s a scary proposition. The laws aren’t going to protect us from that because when the is large enough, there’s always some extreme something at the edge by definition and always a work around technology. I mean, I think most of technology at least interesting stuff is some kind of work around.
Anne Kelly:
But there does need to be some bit of transparent disclosures. For example, Colorado has proposed privacy regulations that would require AI companies to have AI specific transparency in their privacy policies to list high impact decisions and doubt. So those kind of consumer protections, you can write laws, they are enforceable and they do help guide us
Speaker 2:
Forward. Good. I’m scared. I’m glad. Stop fucking on. I’m not totally optimistic it’ll work, but I totally think it’s the work you got to do and
Anne Kelly:
It’s for the consumer to read the terms of service, the terms of use and the privacy policies. How many of us actually read the dolly terms of use? How many of us have read chat GTP?
Speaker 2:
But I haven’t read Instagrams. I haven’t read. It’s like long. In fact, I fed Facebook’s into chat GPT just to shorten it, summarize it, and give me what I should be worried about.
Anne Kelly:
So then it’s the consumer’s responsibility. It’s my choice not to download and utilize TikTok, even though it would probably at some level be helpful, but fun. Yeah, it’s just their terms of use are so beyond. I can’t ethically do that and safely do that. But it becomes also about consumer education, which most people, you’re right, most people don’t care. They don’t care. But having some sort of framework will be helpful. When cars were new, there weren’t seat belts and it took car accidents to decide we needed seat belts.
Speaker 2:
Don’t you think that if they announced today that an iPhone, that this device will a hundred percent kill you from the radiation coming out. We just discovered this today, don’t use your iPhone, it’s killing you. I’d say, what? 30 40% of the country would say, fuck you, I don’t care. Or I don’t believe it. I’m so dependent on it. Let it kill me. I’d rather be dead and have it than not have this device anymore. And I think we get we to these technologies very quickly, and then when they want to take ’em away, it’s like our rights are being fringe somehow.
Anne Kelly:
Wishing that analogy a little bit further. The hypothetical that you po, it’s not whether or not we use a cell phone, but how do we make cell phones safer at that point? If we know that a cell phone causes cancer, then it is the duty to write laws and regulations that force the producers of cell phones to make them in a way that is safe to the consumer. It’s about consumer protection. Again,
Speaker 2:
Then this is about consumer protection, that our government in theory is us trying to keep us safe for people taking advantage of us at some level.
Anne Kelly:
And if something’s inherently unsafe, then that company should have to reconfigure the way that they’re creating that product. So on the larger scope with artificial intelligence, going back to people think it’s going to save the world or destroy the world outside of art, there’s a lot of amazing applications that can help a lot of people and a lot of things that could hurt a lot of people. And in the art context, it’s maybe not as bad as the Terminator, but people have that concern. Artists have that concern that AI is going to take their job because this thing is going to do their job better than they are, which I don’t think we’re there yet.
Speaker 2:
And I have some questions about that happening. I mean, these are just ED theories. I’d like to bounce off of you. I mean, humans are pretty competitive in a way, and so it kind of raises the question like Ruben said, we’re seeing people trying to take that position, but does this use have the possibility of virtuosity is what I’m saying? Can you be a virtuoso at it right now? I don’t know if it’s quite there because we’ve seen this sort of photography. People want to claim that they’re a real expert at this to a level that nobody else has that capability. I mean, I’ll sometimes argue that’s why we have raw files and oh, only an amateur would use SRGB. A lot of these things get bandied around to separate the professionals from the amateurs because digital at first took away all that mystery of the dark room.
Speaker 2:
How are we going to see the point with ai? Because right now the control is really minimal. I think you have the text of things, but artists and humans like to have that specificity. You can’t make that kind of complexity that an artist can. Building metaphors, for example, that run through your image. Think of a Mackey Taylor image for example. And so are we going to have to have AI that does have to have that level of control increase in complexity that is going to give rise to that virtuosity in a certain amount of public to make it interesting to them. Otherwise, it’s sort of like I’m the world expert at tossing coins, who’s going to aim for that title? Whereas I can do things with this tool that nobody else can related back. When Dolly first started, I sat down with a group of friends of mine and I made a prompt about they were their personality in a certain style that I thought fit them.
Speaker 2:
And in relatively short order, I made a couple dozen portraits of each person that they loved that they felt captured their spirit uniquely. I purposely tried to do things in different styles that kind of fit who they were, and it’s a phenomenal ray of kind of cool looking things that they were thrilled with it. Are they arts? I mean, they’re very stylistic and each person felt it was incredibly personal, but that’s a reflection of something that’s called the Barnum Effect, where you can take anything that’s vaguely broadly applied. And that’s not an afternoon, that’s a PT Barnum. This is a psychological thing. The Barnum Effect says, oh, if I’ve got something that’s broad enough, people will read into it. aScope kind of stuff. Yeah, exactly. So it doesn’t take too much to get people to kind of agree with that. The question, are those going to be lasting art, which is always kind a crazy thought art.
Speaker 2:
Yeah, definitely have no problems with that douch shop. But is this going further? I mean, do you see any of those that show real compositional sophistication, sophistication in the use of color that you find really intriguing? They are using some of these images. They feel like it’s cool painting of them. It’s art. If they keep in a alive, if it ossified into their history, maybe they’ll print it out at some point and it will feel like art. Are we the point now where there’s the excitement and enthusiasm that NFTs had a while ago that’s kind of I think kind of fading a lot of people like, oh, change the art world and all of that. And I think it needs the sense that people can feel like they have the potential virtuosity in this medium before they’re going to adopt it as their maybe primary medium. Other than that, I think it might sort of fade. I wish I could predict the future. I think there’s going to be a surge then kind of a dip. People don’t care much anymore, and then the technology is going to change to a whole new level, at least in the visual arts. Well, what
Anne Kelly:
Does everybody think? Do you think in 10, 15, 20 years? It’ll just be like Photoshop.
Speaker 2:
Photoshop already has AI stuff going into it, and in two more years you’re just going to describe stuff. I mean, you’ll be able to grab the reins, but you’re going to describe, get the background out of this. Move the bird off to the left, make it a little tighter. Change your hair, greet. You’re not going to have to have the skills of a Photoshop artist where you have to learn how to select, and you have to learn how to do all the layering. It will make it that much easier, and I think it’s going to be desktop publishing back when that happened. There’ll be a huge proliferation of lousy stuff, opens it up to lots and lots of people who are free now and there’ll be occasional good stuff and important things, but it’s rare. But they wouldn’t have happened at all if the seeds hadn’t been thrown so wide.
Speaker 2:
And some child will do something that will be the next masterpiece and everyone’s going to be, oh my God, and then that’ll be a flashpoint or someone will make a billion dollars on something they did with the AI and it’s going to make everyone go crazy. So yeah, I think that’s a great analogy. I mean, we’ve been kind of talking about two things with, we’ll say the practical useful things. Corporations are going to have a huge drive just so cost and the arts. I think we like things a little bit more messy, harder. We like things to be harder. Oh yeah, that’s where that virtuosity comes from. It’s sort of funny. I mean, we’ve got another little parallel because when Photoshop came out, people that would do spectacular straight photographs would say, no, Photoshop was used on this. Now anybody that’s pretty good with Photoshop is saying, no, AI was used on this. Hi. I mean, I posted an image that I just Photoshop together. I thought, oh, people are going to think this is just an AI creation.
Anne Kelly:
You’re like, no, that was Photoshop.
Speaker 2:
NOIs were harmed in the making of this work. I don’t like you have to put on your caption. This was real. I didn’t Photoshop it or this was Photoshop, but not eight AIed. Ultimately, the metadata gets stripped off of everything and you’re just looking at images, and I have a wilderness photographer friend. He very clearly photoshopped in the auroras over a polar bear, and it’s a gorgeous picture. It’s a gorgeous picture. But he had to say, I didn’t shoot it like this. I mean, I shot the auroras and I shot the polar bear, but I put ’em together. But within one post, people were like, oh my god, this guy’s the photographer. And they’re sharing without that metadata and it’s suddenly is widely seen without any of that information, honest doesn’t mean it isn’t rutted up the world.
Anne Kelly:
Then what do we all think about artists who are potentially using elements of AI and then taking that and applying it to something else? An example would be Maggie Taylor, who’s master of Photoshop, she recently started creating select background elements and then montaging those into her composition. So the entire image isn’t artificial intelligence, but she’s using little elements of it, and I think there’s a lot of room to play with things like that. And as a gallery director and just a lover of art, that’s more intriguing to me personally. I don’t know how you guys all feel about that. And it would be copyrightable, right? All the human elements would be copyrightable so she could exploit and protect the copy and creation of derivative works.
Speaker 2:
She’s the virtuoso that we’re describing, and you want to give people tools. You want them to have every tool they can to create, and that’s fantastic. I just happened to be a purist about definitions of what makes photography, but not what makes arts. I just think that photography has to carve itself out because it’s just a different thing. It’s a different beast than the compositing, artistic magical creations of people like Mann or Maggie Taylor or anybody using Dali and all the other tools. I want them to have the tools. I just think our understandings of virtuosity are going to change,
Anne Kelly:
And I think there’s always going to be a place for fine art. I don’t think it’s going to be supplanted by this. I pulled up a case, there’s a Berlin based digital artist, Julian Van Deacon, and she an AI reproduction using midjourney to recreate another version of one of Vermeer’s works. Then there was a museum in Berlin, I’m sorry I don’t have the name of it, but anyway, they displayed her work and they had 35,000 submissions on an Instagram post about how terrible it was that they were displaying this artwork that was copying Vermeer. So the public I think, is also going to guide this.
Speaker 2:
It was too easy. Personally, I don’t think art should be easy. I think it needs to be something people have to work at and be good at and have craft in it. I know there’s a lot of quickly produced cool pieces of art, but generally speaking, I would say that it needs to take more effort. It needs to be harder. Bar has to be raised because the tools are better. And part of that, to go with Arthur dto, he says a by definition is a status term, and we often don’t talk about the status elements of art. Almost every artwork has some aspect of, and I have my own little formulation of three things, model making the sensate status, and like I said, we don’t talk about the status part that much, but just about everything has the status aspect to it. I mean, if you’re like, say a Japanese ink painter do Sumer, yeah, maybe I do this painting of bamboo in two minutes, but then there’s 18 years or 20 years of study and practice behind all of that. So you need the comm. Yeah.
Anne Kelly:
So what do you guys think about this analogy as art experts, the way that these companies generate and scrape data from the internet of all of these images and copy them and learn from them. How is that different than somebody going to art school and being exposed to all of those different works of art and then being asked to do an oil painting and the style of this person or that person? Because to me it doesn’t feel that different.
Speaker 2:
Well, when postmodernism made its way, we’ll say from France to the US and shook up art schools, part of the notion of that under the guise of death of the author was that everything is derivative in some sense that we’re just reconfiguring all the themes of literature, et cetera. So in some ways I don’t see it as that different. I mean, there’s a lot of history of the tradition of teaching art by imitating masters, so I don’t see any real conflict with that. I mean, Ruben, you’re always good for thinking of something that I haven’t thought of as
Anne Kelly:
A purist. What do you think about that more so? How does that play into this conversation, AI in general? Where do you kind of fit being a purist? Very open to ai.
Speaker 2:
I like tools and I like democratization of tools. I think it’s important for lots of people to have access to stuff, and it opens the playing field up, which lets people grow up learning things and standing on the shoulders of other talented people and doing something innovative. So you have this broad long tail of crap that’s created of imitation and lack of technique and lots of stuff, but it is from that primordial soup that people put it together in ways that no one thought of. They do something in a different way and everyone else goes, oh wow, that’s the moment of creation. I think you’d have to put up with a lot of noise. That’s all. Art is really, I think, fascinating and unique in that it’s highly radical and highly conservative. I mean, artists are jumping onto any new technology they can get their hands on.
Speaker 2:
Yet there’s people using every vintage out of date process. Anything you can imagine. People have used to make art. Somebody is still doing it somewhere and they’re doing retro stuff, and then they combine it with something that is a new medium that didn’t exist, and suddenly something new happens. Like God, I was doing all those daguo types, but now I’ve scanned them into the computer in a way, and you can get something now that you couldn’t have gotten if you hadn’t started from this glass plate with chemicals on it or something. I mean, I love that people do that, and I do think that it’s important and there’s no putting the genie back in the bottle. But I do think certainly as far as photography goes, I think photography is dead. I keep saying that, and long lived photography, just like the king is dead long lived.
Speaker 2:
It’s like what it was, what its purpose was, it’s utilitarian nature. All that stuff is supplanted. You don’t need photography. You don’t need the skills of a photographer to create things anymore. However, if I’m going to use a camera, I’m going to have almost a religious discipline. For me, it’s like a zen art. It’s like calligraphy. It’s like archery where I’m going to do it in a kind of traditional way as a practice, and that’s another way to create stuff where it’s not about out here, it’s about how you do anything is how you do everything. And I’m just going to perfect this simple little thing because nuance is the thing, and I’m going to be constrained. People who are taking pictures with film and with box cameras and even digital cameras and printing them, it’s going to be quaint. But that’s photography in a way, and all this other stuff is another new blossoming cool kind of arts that will be on screens and maybe it gets printed out, but it’s created in very different ways like frescoes and oil painting.
Speaker 2:
It’s just a new thing. It has its own constraints. It’s new own opportunities and its own limitations. I think that the definition is just broadening. I don’t think you can really narrow it that. I mean, the fictional aspect of photography goes back to Bayard right at the beginning, a self-portrait as a drowned man. Dead people do not make their own self-portrait different in that at the time, there was still sort of a sense that when the shadow was frozen on this plate, no matter what fiction you created, it had veracity to it. It was a real thing that was frozen. Even if it’s a scene of fictional scene I created, I don’t think people look at photographs or will look at photographs with that same affectation of authenticity. It’s like a painting. No one looks at paintings and says, oh, I wonder if that really happened, or is that exactly what it looked like? No, it’s impressionistic. It’s when I was feeling about the thing and I can put everything right where I want them. That painting is the metaphor of modern digital photography. It’s not photography. It’s painting in this medium. The medium has broadened its definition.
Anne Kelly:
I love that idea that the medium is expanding its definition.
Speaker 2:
Painters think of their medium is broader than just paint. These days we have drawing classes where people are making vinyl cutouts on the computer and putting them on the walls. I mean, they’re calling that drawing. So I think every medium is trying to expand and colonize new territories like that, hate it or love it. There’s no real answer to that.
Anne Kelly:
Maybe we revisit in the future.
Speaker 2:
Let’s take a look at chat GT four in a couple months and see if we still agree with what we said tonight.
Anne Kelly:
Also, these lawsuits as they move forward. Oh
Speaker 2:
Yeah,
Anne Kelly:
Yeah. Let’s reconvene. But I have a really quick question, everybody. Let’s start with Ed. If you could have a superpower, what would it be?
Speaker 2:
Oh, it’s sort of funny. I was actually thinking about that sort of this week and I’ve decided that I already have a superpower. My superpower is wherever I am is where I’m supposed to be. In some ways it’s like how could I be anywhere else? But it also becomes a reminder to me that, oh, if I’m a little late, don’t freak out or wherever I am, make the most of that experience and the people that are around there. So I said, that’s my superpower to be wherever I’m supposed to be or wherever I’m is where I’m supposed to be.
Anne Kelly:
That’s so funny that you said, ed, do you already have a superpower? I was thinking I already have the superpower of being able to read people’s minds. Sometimes I like it, so probably flying or breathing underwater or being able to transport myself quickly to other places. And how about you, Ruben?
Speaker 2:
I know it’s not flying because I put on a VR headset once, so let me fly, and I almost lost my mind. So that’s out. That’s totally out. As a photographer, I really like being a fly on the wall and the idea of being invisible to see reality without my presence changing. The scene fascinates me a little bit, but even as you were talking, Talia, the instant teleportation is really the thing to make distance not a thing anymore that you can be anywhere. It destroys boundaries and privacy. It does a lot of weird things when distance doesn’t exist, and I’ve always thought a lot about that, so that would be cool.
Anne Kelly:
I have a confession for everyone. That question was suggested to me earlier by chat GPT.
Speaker 2:
Is it going to decide whether we get delivered? Die now.
Anne Kelly:
Wait. I want to hear your prompt. What was your prompt? I just asked for a fun interview question for the end of an interview, and it actually didn’t provide a few options. I was like, that’s it.
Speaker 2:
It’s like mic drop. I got it the first try now, definitely a copyright violation. Video lab did a story on that. Can you imagine for ais having a conversation like we did tonight, even in the future, I really can’t
Anne Kelly:
Heard that they can and will do that.
Speaker 2:
Not as good as ours.
Anne Kelly:
Well, no. I find the article where they pitted two against each other and it became a very confusing conversation. I’ll
Speaker 2:
Try to find that. The Turing test we’re right at that point where you can’t tell. That was always the thing my whole life was that sat out there somewhere that you could tell you were talking to a computer basically and you can’t. You just really can’t. Now that’s new. We’re in a new world and it’s more than NFTs. Palia can tell. She can read minds
Anne Kelly:
And for those watching that aren’t familiar with the Turing test, he came up with that 1950 and the idea was if you could distinguish between humans and artificial intelligence and at the time, no way. Now, maybe next week, I don’t know.
Speaker 2:
We’re right there. I’ve been talking to things that I wasn’t sure and I’ve also listened to Voices of Friends and the question was, is this me or is this my robot day one?
Anne Kelly:
So let’s do this again in a few months when chat GPT-4 has been out a little bit longer. All of this is moving very quickly. I really thank everybody for joining us tonight. Super fun. Thanks
Speaker 2:
For having us. This is cool. Thank you. Thank you. You people are wonderful.
Anne Kelly:
You guys are awesome. Fun conversation. Let’s do it again. Totally Goodnight everybody. Thanks for watching Art in the Raw conversations with creative people. I hope you enjoyed the conversation and learned a little something too. Stay tuned for part two. Next time we’re going to talk about music and if you enjoyed the conversation, please tell like-minded friends, like comments, subscribe, leave a comment below, ring the little bell, all those good YouTube things so that we can spread to other people. Have a good night, y’all.
