Presentation 2 // AI as Creative Collaborator
Video Description
This is the Second in a series of presentations by American Techspressionist artists at the new Makerspace in the American Corner in Qarshi, Uzbekistan. Three artists presented imaginative creative works via Zoom, focusing on the use of Artificial Intelligence in creating Digital Art. These American artists are part of the global Techspressionism network. @Techspressionism
Synopsis
Click a time code to jump to that point in the video, or an artist’s name to learn more about them.
Presentation 2, "AI as Creative Collaborator," is hosted by Cynthia Beth Rubin and features Michael Pierre Price, Verneda Lights and Gregory Patrick Garvey. Rubin introduces the session and the presenters: Price, a former game designer with a background in physics and astronomy▶ 4:55, Lights, whom she calls a polymath artist and STEAM innovator▶ 11:52, and Garvey, a professor of game design▶ 19:28. The panel's discussion starts with when each of them began working with AI▶ 26:55. They talk about how far image generation has come in five years▶ 30:34, early experiments with code and video frames at the MIT Media Lab▶ 33:55, the arrival of large language models in 2022▶ 37:17 and using color and abstraction for stress management and calm▶ 40:37. Other topics include Abstract Expressionism as an inspiration▶ 47:05, whether AI images lose meaning as they get more polished▶ 50:08, AI in architecture and 3D modeling▶ 59:59, whether we are living in a simulation▶ 1:03:07 and why the same prompt never gives the same result twice▶ 1:06:12.
This summary was written with AI assistance from the recording’s transcript and reviewed by the archive’s editor. The timestamps link to the moments discussed. Please check details against the video.
Participants (4)
Watch on YouTube // // Download transcript as
Cite this session
Hello Uzbekistan Presentations, “AI as Creative Collaborator,” recorded February 26, 2026. Techspressionism Video Archive. [DOI pending Zenodo deposit]
cha :)
Hi, welcome to our Techspressionist session on artificial intelligence as creative collaborator. I'm Cynthia Beth Rubin. I am one of the people organizing these sessions which grew out of the exhibit we had in our Urgench which I'll tell you about in a minute. Just each of you who are presenting today just say hello before I start screen sharing. So, Michael, hello. And good morning, evening everyone. Okay. And, Verneda. Yes. Good evening, everybody. Good to be here. And Greg, good evening. And it's great to be joining you all. Okay. And Cha who's moderating the question period. Good evening everybody. I'm really honored to be here and sharing this time with you. Thank you. Okay. And now we have a few other people online who will be presenting or have presented in the past. So we're a really nice group.
cha :)
And I'm just going to jump right in because in 30 minutes we're going to stop for a minute. So, we're really happy to see everybody here. We are part of a group called Techspressionism and today we're presenting AI artificial intelligence as a creative collaborator. For those of you who have missed our definition of what Techspressionism is, it's a fairly new term. Just go back for a second.
cha :)
And it's we have a group of Techspressionist artists and it's just what it sounds like taking expressionism and technology and putting them together. This is our website and the website for our show. You are welcome to join this group. We have virtual co-working and we had an exhibition at the museum of contemporary art in Urgench. So that was one of what brought us together. Here we are at the opening of the big exhibit. And here's Shirin Tashova who was the director. We have these weekly we're doing sessions with your group every two weeks. So please come to our upcoming sessions and we are going to jump right in. I'm going fast to give us time to the presenters.
cha :)
The presenters are mostly going to talk about their concepts. They will give you some ideas of tools. Here are some common AI software. And you have Copilot is a Microsoft, so it's on your computers. It's really easy to use where you open it up copilot and you have some credits. Credits are how many images you can make. You will hear many of the artists say they take their work into Photoshop, but on your computers you have Gimp and Microsoft Paint. And you will be able to do some of the techniques, the same techniques that people do refer to in Photoshop using Gimp. It's just a same idea, slightly different interface. And then you can make layers in Microsoft Paint and also paint on things. So that just to give you an idea that when people talk about their tools, you don't have to get exactly the same tools that they have.
cha :)
We would we have a handout, I believe, with a lot of terms and people will be defining their terms as they go along. These are very important things. You'll hear prompt over and over again and it's what you tell the AI that you want to do. And then people also say iterative because creative users don't just use the first thing that came out especially when we're talking about artificial intelligence and AI. The first thing is almost never what you want. So that term you'll be hearing a lot. So these are the artists.
cha :)
I just introduced them very quickly and when we finish Cha who is a creative artist himself will be presenting the questions we'll be coordinating the questions. So our first presenter is Michael
Michael Pierre Price
Pierre Price. He has a background in physics and astronomy. He worked as a game designer for 30 years. Games that you probably know, Dungeons and Dragons, amazing number of games that he worked on over the years. And for 15 years, or more than 15 years now, he's been doing creative work as a computer-based artist, and he's one of our very creative AI users. So here's his website and here is Michael. Thank you. I'm inspired by the profound nature of the universe and existence which motivates me to explore the intersection of physics, neuroscience and consciousness and spirituality. I concentrate much of my efforts into making my artwork around ideas like the nature of reality, the fractal quality of the universe, and the sacredness of existence. I want my artwork to challenge the viewer to contemplate the universe and existence in ways they may have never done before.
Michael Pierre Price
Much of my art is highly symbolic in order to open gateways to imagination. My artwork oftentimes isn't so much about giving answers to the great mysteries of the universe. It is as it is about asking questions and making the viewer think. And I use AI as a brainstorming partner. I have fashioned my own way of working with AI that's very different than the norm.
Michael Pierre Price
And it isn't so much that I have specific visuals in mind to begin with, but I have big ideas that I'm trying to explore. And through my use of AI that I've developed over the last several years, that presents to me many unexpected connections which I find very interesting and challenging to my thought processes. So the normal way that AI is fashioned and designed is for an artist to have something in mind. So they have an idea of a visual they want to create. And the approach that I've identified this as is what I call the artist director where the AI is a servant to the artist.
Michael Pierre Price
And so the artist needs to have a clear, precise, and focused set of directions, which we call the prompt, that will allow the AI to create images that align with the artist's intent. And that's a very direct way to work with it, which I don't use. And I want to explain to you what I what my new process is. And I call it the artist explorer where I use the AI as a collaborator. And in this approach, I have an idea in mind, but I don't know what the visual is going to be. And it's through this iterative process of brainstorming that I finally get to something that looks interesting to me enough that could become a piece of art. And so I use the AI with this idea of using the richness of language to actually subvert the AI in ways. I use ambiguity, paradox, questions to look for novel ideas.
Michael Pierre Price
And so for me, the AI is a brainstorming feedback loop that I use to make images that will emerge into final pieces of art. And this is the approach that I used to create the three works of art that are on the artist reel. And I was exploring the idea of consciousness within all life. And this first piece, this first artwork is about the rudimentary existence of consciousness within the most basic forms of life and how that's existed now for hundreds of millions of years. And this next image was looking at higher life forms that have selfawareness that they're aware of their existence and the world around them. And I was exploring that with this work. And then the final third image was looking at the life forms that can actually dream like human beings and what dreams mean to us. Being able to take us outside of our normal reality and explore vistas we would never be able to see in normal life.
Michael Pierre Price
My process created for more than 1,400 images that I tried to look at of which 331 of them I found interesting enough to save and from those came the three final artworks for the artist reel. I Oops. Not sure what happened there. Okay. Through my process, I actually had 14 prompts that I worked with. The three that are highlighted in yellow were the ones that I looked at early on that I spent the most time and created several hundred works that I thought could be interesting. And then the final two were the two prompts that I landed on about the non- euclidean existence of consciousness. Here's several examples of images that got created along the way. Here are the two prompts that created those final three works of art. And in midjourney, I'm able to use existing artwork that can influence the text of the prompts to create the images that potentially could be final works of art along the way.
Michael Pierre Price
And here's the software that I use to create my works. The four that are highlighted in blue are the actual ones that I use to create the three final works of art. And then here's my hardware from my computer to my external hard drives that I save most of my imagery on. And then finally, the printer that I use to print my work. And I wanted to give you here just a quick overview of the three images of that were in the show. And so I want to thank everybody for listening to my presentation. And I would encourage you to come visit my website. You can learn more about me, what I've done, what I am doing and my artwork. So, thank you very much. Great. Thank you, Michael. That was wonderful. Really great. Thank you. So, our next artist is Verneda. She is a
Verneda Lights
Polymath artist. She's a steam innovator. That's steam is a term that we use for people who include science, technology, electronics, art, and I can't remember what the M is math. Math. Oh, thank you. Oh, she is actually a doctor. She started out in medicine. Before medicine. Of course, she had a liberal arts degree and then in the year 2000, she began to learn computers and switch over to digital art and thinks about color as healing, especially after our incidents in 911. And she has exhibited widely since 2003. She's wildly and wide widely, excuse me. And in fact she was here where I live in New Haven for a Yale symposium at Yale University and presented in exhibits really all over the world. She's been exploring all kinds of different creative methods which she will tell you about and her philosophy includes art as medicine.
Verneda Lights
So she draws on her background as a doctor in creating using creativity and creating new world. So hello my name is Verneda Lights. I'm an artist who uses technology to explore ideas, culture, and human connection. In my practice, generative AI is not a shortcut, but a research partner. It's a way to listen across cultures and across time.
Verneda Lights
Caterpillar Dreams Stand Up is a playful vision of a being that is part human and part butterfly created with Night Cafe AI. I refined it in Photoshop. It was presented at the Yale Ultraspace Symposium and featured in the Techspressionism reel in Urgench. Now, the first step in creating an image with AI is to write a prompt, a short description of what the picture should look and feel like. I also tell the AI what I don't want, like extra fingers or arms. These mistakes are called hallucinations. For this presentation, I collaborated with ChatGPT to write the prompts.
Verneda Lights
This is the actual prompt that I gave to midjourney. I described Qarshi as a silk road meeting place full of traders and textiles. I also carefully guided the clothing, fabrics, and architecture so they stayed historically accurate. Finally, I added a negative prompt to prevent mistakes like extra fingers or distorted faces. Here I ask midjourney to imagine a series of digital paintings called silk road. In a way today we are also meeting on a new silk road the internet where ideas travel instead of caravans. Looking at the four images which one would you choose?
Verneda Lights
This is the silk road image that I chose. I wanted the scene to feel timeless, so I checked for modern details like zippers, high heels, or handbags. The colors also felt muted and the light a bit ordinary, so I refined them in Photoshop and Lightroom. After Midjourney created the preliminary image, my work as an artist continued. I use Photoshop to refine the color, adjusting exposure, contrast, saturation, and sharpness. Then I use Lightroom to shape the light for a natural feel.
Verneda Lights
This is the prompt ChatGPT and I created for the second image about women in Qarshi who embroider. I focused on rhythm, layered color, and symbolic patterns to reflect living culture. I wanted the clothing and fabrics to be authentic. Studying these four images, I noticed a tool that I hadn't seen before. And I researched it to see whether or not it was an hallucination. It turns out that it's called a tambour hook or a yerma.
Verneda Lights
So this small hook allows you to do embroidery in a way that looks like drawing. This is the uncorrected folk art image from midjourney. Even though AI helps create beautiful images, artists still have a responsibility to verify their work. But I researched the embroidery tool to make sure it was actually an instrument being used in the culture and I also checked for anatomical errors.
Verneda Lights
So while reviewing the folk art image, I noticed the index finger on the first girl's left hand looked distorted because of shadows. I corrected this in Photoshop so the anatomy would appear natural. This is the corrected finger. And you can see how subtle differences can help to keep the realism intact.
Verneda Lights
Here I noticed the thumb looked a bit distorted. So I brought the image into Photoshop and repaired it using painterly brush work like regular painters do to restore a natural hand. Here the right thumb has been corrected. Some minor adjustments were also made on the left hand. After refining the color and anatomy in Photoshop, I moved the image into Lightroom to adjust light, the warmth, and the atmosphere. This is the final image I call folk art. And I hope that you find it to be an accurate depiction of the textiles and people of your culture. And here you can see some of the embroidery techniques that are used in Uzbekistan being imported into my business logo.
Verneda Lights
And I would leave you with this thought that tradition and technology can grow together. Thank you very much. Great. Thank you so much Verneda. That was really wonderful. So our next
Gregory Patrick Garvey
Presenter is Gregory Patrick Garvey who lives in Hampton, Connecticut. He is professor of game design and development at Quinnipiac University in Hampton. He founded the game design program. He started it and he also had served as chair in the department of visual and performing arts at the university. Previously he had been the chair of the department of design art at Concordia in Montreal, Quebec. He has been an associate artist with the digital media center at Yale University and he was a fellow at the very prestigious high powered important center for advanced visual studies at MIT. He is a very active exhibiting artist. He has presented his artwork all over the world.
Gregory Patrick Garvey
You see a list of countries there and he will tell you himself about exhibiting at CAMUZ in Uzbekistan. So here you go Greg. Thank you Cynthia and it's an honor to be joining you today here in lovely Hamden. And we'll move on to the next slide. So I participated along with a number of the other artists in the Hello Uzbekistan at the contemporary art museum I. Well the exhibit took place in Uzbekistan in central Asia. It actually began for me in Kuriyama Fukushima Prefecture in Japan at the Shinto shrine, Kaiseisan Daijingu shrine. And I took photos like this of angular branching of pine trees on the grounds of the shrine. And I seek to explore the aesthetic of glitch art which embraces digital or analog errors, signal interference, image breakup, and distortion. I uploaded photos of trees into Midjourney, and I wrote prompts to distort the source images increasingly towards abstraction.
Gregory Patrick Garvey
Here you can still see the remnants of the branches from the original photos. And this is the prompt that I repeatedly use. Cable TV glitch screen pixelated sharp edges break up bright saturated colors emphasize compositional center vertical stacking angular zigzag vertical branching vertical glitch breakup colorize invert complimentary colors pixels and the these are examples of repeatedly uploading and repeating the prompt again where the output becomes comes the input for the next image. And this prompt shows the multiple generations and the result and where I select or curate the images from the many that are produced by the AI. And I begin to tease out and emphasize vertical and horizontal elements as you can see here along with complimentary colors very pixelated and hard edges. So, and you can see another variation here as I explored the space of possibility.
Gregory Patrick Garvey
So, the Kaiseisan Daijingu Ki series is a continuation of a longtime interest in what we call today glitch art. And this that is seen in an early portrait from 1988 called Ted and Liza using the true vision image processing software. That's before Photoshop. At the exhibit, students who attended delight in taking photos. This is the local artist Qodirov Soliy and Uzbek artists in one of his paintings who joined us. And this is my artwork designed to lead the eye in an unexpected ways through gaps in a predictable grid arrangement in the exhibit. And then I created a video and I used a very simple prompt slowly evolving glitch distortion 3D rotations based on the generated images.
Gregory Patrick Garvey
The music is by Emmy award-winning sound designer, composer Serge Ossorguine. For another AI generated work, I used a simple prompt, a distinguished professor gives a lecture and the following slide shows the results of repeated iterations. Weights and biases have a specific technical meaning in machine learning for training AI but become a metaphor for the inherent bias built into discussion models. And finally, these terms play an important role in how I theoretically think about the use and abuse of AI generated images. Please refer to the handout. Thank you so much. Okay. Sorry that went a little fast at the end.
Gregory Patrick Garvey
Let's see if people didn't get Greg's website will we have this I think you have the handout that has it. So thank you so much Greg that was really great really comprehensive. I will introduce you in a minute to our moderator Cha and I want to remind you of our upcoming sessions. We are doing
Announcements
These every two weeks. All different ways of how to create imagery with computers and we want to thank American Makerspace for having us here and also the contemporary art museum of Uzbekistan and all the contributors for bringing us together. So Cha if you want to take over with a few general
Discussion
Questions for people. Good evening everybody. So glad to be here with you. And very honored to participate in this. I guess just a quick we don't have much time but a quick question. We'll start with Verneda and I'd like to know more about how you began your journey with AI. When did that when and why did that come into your practice? Well, I wanted to be able to explore storytelling in a way that I had not done before. So I saw the ascendancy of AI and it seemed to me to be a gateway to a world of possibilities and so I started experimenting and wanted to see how the AI would capture various concepts and when I interact with AI to create artwork I use very informal prompts. I mean today I shared the ChatGPT version which is very structured, very formal and it gives definite parameters to the AI as to what to do and what not to do.
Discussion
But a lot of times I give leeway to the AI's imagination and interact with it based on that. And when I saw the storytelling potential I decided that was something I wanted to do more of and also I wanted to explore different cultural experiences across the planet and AI seemed early on a bit limited in that regard. So I wanted to have my involvement be a training experience for it.
Discussion
Great. I think we're kind of reaching our pause moment here. So okay, let's do that. Yeah, I believe that there's water and snacks for people and for those of you who are breaking fast and we'll come back you can signal us when you're ready. Just before we do, I just have a quick question. How many people here have used AI in the group in Uzbekistan?
Discussion
Have any of you? You're muted. Okay. I'm just curious if you raise your hand. Thank you very much. I use it a lot. Okay. Well, we'll Discussion that after breaking. I don't know. We'll just signal us when you're ready to come back. We'll be here. So chat up to you now. Welcome back everybody. Michael, I'd like to ask the same question of you how you started your journey with AI and then in addition to that I'm curious about the philosophical question you're asking more than one but what is it?
Discussion
I think for me it started just a little over five years ago. And it's rather mind-boggling to step back five years ago to see where AI and image generation was at that point where it is today. There's been so much evolution and change in what's possible. And so I relied a lot on my game design background and running software companies for making games. Brainstorming, coming up with ideas is a really important aspect for creating anything. And when you can do it with other people, what you find is that the ideas that get generated are greater than what any one individual could do. And so for me, I saw AI as a possibility of collaborating with a nonhuman entity and brainstorming the big ideas that interest me as an artist. Those things like what is existence? What's the universe about from the smallest to the largest?
Discussion
And so for me, that's what began my exploration as looking at this as a partner that could collaborate in helping me find steps along the way to create the kinds of visual imageries with symbology and such that would help articulate the thoughts that I have about existence, the sacredness of it all, the fractal quality of the universe and things like that. And it's definitely been an ongoing journey and it's still one that I find interesting.
Discussion
That's interesting to try and answer a question with your artwork. I'm wondering if anybody at the at the Makerspace in the process of making art, are you are you trying to answer a question or have you ever tried to answer a question with your art? No. Well, now you may know that's possible. A way to use art as a way to learn more about yourself or the world. So, now a question for you, Greg. I'd like to know more about the start of your journey using AI and then also why the glitch? Well, and I'll start with the glitch first. When I first experienced paint systems, I saw the possibility of, and this was actually going back to like 1981 at the architecture machine group at MIT, which is the precursor
Discussion
To the media lab. I worked with a dancer and I frame grabbed shots from videos and I began to see the possibility of pixelation and distortion. And this was written the code that I developed in was PL1 and this allowed these interesting breakups. At the time glitch was not part of my vocabulary. So it actually goes back to the very nature of the medium and I thought about with what is it about working with digital images which has what's its inherent nature and much of say contemporary or modernist impulse is to explore the nature of the medium what the medium can or cannot not do. And so I saw pixels as the fundamental element of the digital image and that you can manipulate those pixels. You can make them more prominent, distort them, play with them as much as you might push paint around with either a brush or a squeegee or a spatula.
Discussion
In regard to AI, I became interested, I had taken a course in the lisp programming language and one of the assignments was to write a sentence generator and this is long before large language models and at the time neural nets were long discarded by there are different camps in the AI community and so I wrote this basic sentence generator and then I modified it to be become what I called the image parser so that it could put together images in different perhaps discovering a semantic relationship between images and then I sort of went away from that but I returned to the idea of working with AI with a work an installation called the automatic confession machine, a Catholic touring test. And there I began to play with the idea of chatbots. And in the end, the version that I then exhibited that would it caught some media attention was written up in Wired magazine, even National Geographic.
Discussion
It used a more scripted approach, but it was an interactive experience which allowed you to explore the theology of Catholicism and the ritual of confession at the same time raising these important questions about the intrusion
Discussion
Of technology into our very personal lives. And a kind of surrender to the epiphany of digital transfiguration if you will. So then as large language models became very became back in 2022 this marriage between language generation and image generation seemed like a very natural step to begin to explore. So that brings us up to today and also I'm keeping with the idea of exploring the very nature of the medium. The medium is pixels, color and so forth. Well, thank you Lisp. That's one of the early or maybe the first AI languages, I think. Yeah. Yeah. Well, so I do we have any computer scientists in the in the audience at the makerspace? Is anybody studying computer science? I think almost everybody here are artists and have studied computer science or at least mathematics. Which is interesting.
Discussion
Yeah. Yeah. Yeah. Basically, he's a friend. He doesn't know English yet. So he learns architecture. So basically he do need temperature. Yeah. Great. Okay. I have a Verneda I have another question for you and this is I'd like to know more about your art as medicine practice. Yes. Well, I started working with digital art in 2000 and then in 2001 911 happened and that really changed my world. I was transitioning out of a profession as a clinician in internal medicine and running a practice and whatnot. And I had not yet decided what my next move would be. I knew I needed to develop new skills in order to best use my experience with the world and to be useful in general to my community. And 9/11 really changed the world. You know the advent of terrorism and destruction, mass destruction was horrifying.
Discussion
So I needed to de-stress from that and I noted that whenever I was working with the computer and using color and I was using MS Paint at the time and I was fascinated by the wide variety of colors that you had access to in such a small amount of space.
Discussion
I was exploring this and I noticed that certain colors would create a relaxation effect. So I started using the color and MS paint for stress management, color therapy and then I started to shape things and do abstractions and from there developing textiles and whatnot. So I embraced this and have continued it throughout the decades when there's moments of stress where I need clarity where I need to work through problems. The art is a way of offering a buffer to explore possibilities and solutions in ways that are non-threatening and in ways that are deeply relaxing and then you learn also a lot about yourself in the process of experimenting with art.
Discussion
So I noticed that the art was doing what we used medicine to do when I practice you know and so I was very glad to come to that understanding that art is a medicine because it's holistic and you have the lowest level of threat that is adverse reactions you don't have to worry about allergies. You don't have to worry about drug interactions or anything like that. You sit down in front of a computer and you just go with it.
Discussion
Well, thank you. So, what if with our primary colors, how do they affect us? Like just the basic red, blue, and yellow. Yeah. Well, red in general makes you very it goes with aggression and energy. High energy. Blue is calming. You know, yellow I'm not so aware of what yellow does, but I know that there are architects and people who design interior spaces who gravitate towards certain colors. In psychiatric wards in some healing spaces for people with mental disorders u they might use blue the color blue on the wall because of the calming effect that it has. So color therapy is a very real thing, you know. And people when they who are into fashion will often choose colors based on how they feel. And some people will even use colors to boost their mood or to help them to feel stronger and empowered in whatever space they are about to enter.
Discussion
So, it's something I think that needs to be paid attention to, you know, because we live in high anxiety times, you know, from nuclear armaments on down. So, I think that art is a gateway for calm and learning. And healing and healing. Yes. Healing. Absolutely. You know the good part of medicine is when well physician treat but God heals you know but the healing is a very important component of medicine and I think it is also for art. I mean, when I was a caregiver for my parents who have, you know, now who are now deceased, I used art to help me grieve their loss, you know, not just document in a photo book or, you know, family album, but I started using the pictures that I took of them while we traveled that journey of caregiving. And then after they passed away, I was I started to elaborate on those portraits to create a secondary and a tertiary and a quantinary level of insight.
Discussion
So it's a multi-dimensional portrait adventure that I went on that helped me to grieve them and that to me was very healing. But thank you, Verneda. I have a question for Michael. I'm wondering if there are some artists who are influencing your work. I'm wondering who those who those are. Oh gosh, that's a really good question.
Discussion
I'd say there is a multitude of them. The ones that are most prominent for me that I can name Picasso for sure primarily because he was a real innovator and to be in a world that's changing like it was around the early 1900s with science changes going on economic world to create vast vastly different modes of seeing the world like cubism to me is very inspirational.
Discussion
Dali with his fantastical worlds of surrealism and the other surrealists because I find surrealism a mode for me that's very effective in trying to portray the world outside of the normal ways in
Discussion
Which we see things. And a little more modern. Jackson Pollock for me is very inspirational as well. And the abstract expressionists were looking at creating artwork that wasn't about concrete images, but about the feelings that they had inside of them using body movement and such to create works of art. I remember the first Pollock that I saw at the Chicago Museum of Art. It just blew me away. I was like 11 or 12 years old and it was this monumental piece of art that I just saw as lyrical in ways and it just profoundly affected me. So just to name a few, those would be the ones that come to mind the quickest. All right. Thank you, Michael. Then I've got a question.
Discussion
For Greg and I know you're asking some questions with your art. So I'm wondering if you'd like to talk about some of those questions that you're asking. Whether it's AI slop or the concept of the ouroboros up to you. Well, I'll start with the ouroboros. It is a limitation of the current training of AI that it is the major models out there like anthropic or OpenAI have scraped the entire internet and now they are generating output that is used as input. So AI is being trained on itself and that is this circular loop and I think it when we consider what is called AI slop which is how easy it is a really good example of AI slop is Sora released the app for phones that is released by OpenAI and essentially you would say it's a meme generator.
Discussion
Sam Alman, the CEO of OpenAI, is trying to compete with Tik Tok. And when you look at those videos, they're basically visual jokes. Many are surreal. There's lots of different memes like the baby that runs away from its mother. Animals go crashing through buildings and so forth. And so there's a kind of flooding the zone we can invoke anyway and they're low quality but
Discussion
Photorealistic or cinematic images but they're almost meaningless. And so it's a kind of crushing of information, a kind of entropy, a kind of like Claude Shannon's notion of information theory that there is low information and so AI slop is this kind of diminution of information. Now a different aspect that I'm pursuing in my artwork is a kind of trope of some 20th century art and even yeah which is the removing the hand of the artist from the work. So many people you can often recognize the work of an artist almost instantly like if you see a Salvador Dali you recognize his style and even the manner in which he painted. Or another example might be Rembrandt or France Halls.
Discussion
You even see a portion you can recognize the gesture the way the paint was handled whereas there's a separate you know Ellsworth Kelly might be an example of an artist where the gesture is removed where he makes a single canvas of a single color and so the my use of generative AI is in a sense removing me my individual gesture but it yet at the same time I am selecting and curating the output. So, it nonetheless reflects what I'm looking for and how I try to nudge it to create an interesting compositional result that then makes in effect comments on the very nature of the medium. So, I'm kind of pursuing multiple threads. And there's a transformation of information.
Discussion
So there's an input and there's an output. Kind of the very simple model of symbolic manipulation which is actually a contrasting theory of AI which is very different from the developments I think it was Rosenblatt who developed the first perceptron which became later the basis for artificial neural nets and machine learning did I answer your question. You did. Thank you, Greg. I just have a quick question for the makerspace. I'd like to know if you have favorite artists, could you share some of the artists that you like, the names of some of the artists that you like?
Discussion
Architecture loves Rembrandt loves US. I'm getting the echo. I heard Rembrandt. Yeah, we got Rembrandt. Thank you. One microphone. Yeah, we're just about out of we're I think you have to leave in a few minutes. So, results my friend just well I know some works of Leonardo Da Vinci and Picasso but I'm not imperfect at this pain for painful.
Discussion
All right. So, do we are we wrapping up now? I think we're pretty much wrapping up. I'm just curious to know how many people have looked at any of the AI software. Have you tried C-Pilot or anything else? Yes. Firstly, sorry for echo. We like didn't solve this problem totally. Yeah, we use mostly use ChatGPT image generator. It's like have limitations but sometimes I was using it for memes or like some seems like in Instagram when you like about yourself. Yeah, it's like under popular use.
Discussion
Well, thanks to Cha for preparing some great questions and getting this discussion going and to all the presenters. We hope that you'll come back in two weeks for our next session, which I lost the slide that says what it is. But anyway and if do people have any more questions of the group before we end.
Discussion
Okay. So, no, we haven't sent you. The next one is, I just had that on there for a second, but the next one should be showing. I don't know which screen I'm sharing right now, but how technology changes artists interactions. And I'm going to stop the share. Now we see only you. You see only me. You didn't see the next.
Discussion
You didn't see that slide. Oh, well, you said about technologies, right? The next one has to do with culture and how technology changes the way we deal with culture. All of these overlap. I mean every artist that you will see actually pretty much does what we had paint last time. People use paint programs. Many artists are using AI. The idea of looking at cultural sources that concept came up pretty much in what everybody presented today. But next time that's going to be the main focus. And so we hope that you'll come back and encourage your friends to come back and go to people's websites. I think you have a print you have some printouts with some ideas. Yeah. And we tried to give you some terms, some vocabulary and definitions this time because so many specialized terms in AI if you're learning English and you're using Google to read about different ways of doing things and they're still using terms you don't understand easily.
Discussion
These are terms that are also new for native English speakers. It's a new world. So, it has new vocabulary. Yeah, it was lots of hard new words for me today in the meeting, but it's really interesting. Thank you for your Yes. Do you have any more questions about any of the words the terms? One of my friends he asked like he interesting in architecture. So can he use the AI for his like studies in architecture? Which oh architecture that's interesting have to write to my architect friends and ask how much they use ChatGPT that's interesting because we our group is
Discussion
All people doing fine art so but we all know architects so it's a good question. Yeah, I would say yes definitely to that with using AI for 3D modeling and even architectural plans. In fact, with architecture, we probably have always been kind of on the forefront of technology in the arts with the CAD machines and yeah, Greg. Yeah, chat is something that you could definitely use in studying architecture because it studies patterns. You know, AI is good at patterns. And so, pattern is a very distinct modality that we see going on in architecture. You know, different ways that buildings look and the different ways that architects conceive of space.
Discussion
So ChatGPT I think should be able to you know with this image capability be able to help architects imagine new spaces for oh I'm sorry for living yeah go ahead I was just going to add that there are new tools coming online to help generate 3D one is meshy which can take a drawing a or I've used images from Midjourney and it creates a full 3D polygonal mesh and the texturing and the shading and the lighting and then can rig the character and also provide a walk cycle and simple movement. Google has developed VEO which can generate worlds in real time and you can explore a three-dimensional world and then more recent Yann LeCun who was formerly at Meta left because he feels that LL large language models are fundamentally flawed because they are basically sophisticated autocomplete is generating AI that can create true interactive worlds with physics much like game engines do.
Discussion
Game engines like the Unity game engine on the Unreal Engine. And this is a very significant development and I suggest that this could be the first step in Nick Bostrom's speculation that we are living in a simulation which he has calculated a certain
Discussion
Probability based on a future civilization simulating past worlds. So are we living in a simulation? Okay. So with that I think they have to leave. Their space is closing. So thank you if you want to contact us by email. I think you have people's websites. Yeah. Right. Yeah. And you can find contact there. And for people in our group, I sent you an invite that we can do a little post if you're still awake and you want to talk a little bit, a different Zoom link that we could hang out for a little bit. But I think it's time for them to close and we have a little time and he said like my friend about last question. What about like problem with AI that he likes new things? I mean basically he has like database so everything he creates he just copying so he has like problem that he doesn't create nothing new so it's like just copying something so the question is how do you make it new and not just copy what already exists certainly so I think There's the nature of large language models is specific using a variety of mathematical techniques and which in some ways just comes down to multiplication.
Discussion
But near the end with the sentence generation, they're using Markov chains, which is a way of predicting the next word based on the probabilities from previous words. The image generators that we're using use what is called a discussion model. And diffusion analyzes source images which basically the builders of the diffusion model hired a large number in this case documented Kenyans paid them low wages to look at images and identify the images with written descriptions.
Discussion
Those images then are decomposed into noise. And then that process of going from noise to a generated image is grabbing little probabilities of images from millions of images. So almost by the very nature of the technology you will never get exactly the same image. So it is almost always
Discussion
Something different and new. You can try to coax it by the prompt to try to replicate but you won't get something that is exact. Yes. I to that I would add I gave you some prompts on a handout and I also gave you the prompts that I used to create the images that I showed you in my presentation and I urge you to take those prompts and put them in ChatGPT and see how what you're given as a result of using those prompts whether or not there's any similarity. Between what you get and what you saw in the presentation here today that I gave you. And also you can try the ones that I did not use for image generation to see what you can get. And you can change the prompt around any way you want to see what types of innovations you can create. So I would urge you to do that.
Discussion
Yeah, great. So, that was that's a great question and I think Michael also showed us when he said how many he had to make thousands, right? Yeah. And just one really quick point is that if you look if you look at this as a technology like early on quick digital cameras if everybody is just using it in the same way. You may have different environments that you're photographing, but they're all there's going to be a similarity because the intent the way that it's being used falls into a very narrow bandwidth of what's possible. And I think for me as an innovator not only in game design but my educational background I look to try to do things differently than what the intent is and I was trained to break software early on.
Discussion
So the ideas behind how you use the software can also radically alter the end output that you get as well. Even within the limitations that Greg was talking about in terms of how large language models work and discussion models work, you can still, as he mentioned, coax the AI in ways that most other people may not be looking to use it. And that and I think that's a really important distinction between artists and people who are just generating images.
Discussion
There's a vast difference between that and I think that's where aesthetic qualities and intent and artistic aesthetics really come into play. So not an easy answer to give you but the possibilities definitely are there even within the limitations that are inherently built in. Okay, great. Okay, thanks everybody. I think we'll see you in two weeks. Thank you.