Interview with David R. Smith – Stone Age Time Machine

DRS: It’s nice to be here. I appreciate the opportunity to talk about what I’ve been doing lately on the music front.

DRS: It is a “band” only in the sense that music in the past and even today is centered around bands. So a “band” and an “artist” and of course a “record label” are the core components in the old world. I should say also in the current world, but things are changing.

DRS: Yes, exactly. So to answer your question, Stone Age Time Machine is my artist name. That's my "musical" self. If you want to look up and find my music, the music I’ve been making recently using AI, as opposed to typical recording processes, then you would search on that.

DRS: Thanks.

DRS. Well, yes, there’s a lot to talk about. Maybe some background is in order. My generative AI journey started years ago with Midjourney 4 and Stable Diffusion. As soon as I saw what those platforms could do I was hooked.

DRS: I went to art school for a while, and yes, I’ve always had an artistic interest. My mother was an artist, she painted watercolors. Still does, actually. But I don’t know where my interest in art comes from exactly. Maybe from a past life!

DRS: I was a ceramics major.

DRS: Perhaps on the surface it appears that way. Ceramics is what my professor always called the study of “materials and techniques.” If you think about it, ceramics has so many steps: getting the materials out of the earth, preparing them, grinding them, mixing them, and then the clay must be shaped. There are so many ways to do that. Next comes drying and first fire (bisque fire) and then glazing and other kinds of decoration, and then high firing. So many steps. I have always been attracted to process and in art school I explored other kinds of processes, like Xerox printing and lithography, and photographic process. All of those things have step-wise and iterative phases. And that is true of digital graphic arts. I think generative AI also has many phases. For example, in order to make a model, it has to be trained. That’s a long process, and involves data collection and so on. Then software has to be constructed to use the model, and then finally someone can actually generate an image.

DRS: Yes exactly. But luckily most of that is done by specialists. As a consumer of those models, I don’t worry about training too much. But some artists are doing that. It’s a way to get results nobody else has.

DRS: Definitely.

DRS: I know. That was depressing at first. And puzzling. But I do have a theory about that. At first, when I saw how fun and addictive generating images was, I thought everybody would be into it and I could even sell what I was generating, because to me, it looked really cool, and also it was a newer thing. But that’s not what happened.

DRS: It’s not quite so sharp a divide as portrayed in the popular media. Serious artists who understand something about technology are using AI, and professional graphic designers have had to at least have a look. It would be career suicide not to. But I think I can explain what has happened. You see, there are different ways of looking at the world—different ways of understanding and reacting to the world.

DRS: In my system there are at least five ways people approach the world around them: thinking or using reasoning about the world, judging, which is a kind of weighing or comparison, feeling (or emoting) about people and things, intuiting, which is a kind of additional sixth sense; and then finally kinetics, or movement. So people who are traditional artists, for example someone who paints, someone who uses oils to make a canvas, like those in a museum—that person is primarily feeling. Color is used to represent emotion. I’m not sure if that’s obvious, but color and emotion are basically the same thing. There is also an element of form, of balance, in paintings. If you look at something by Cezanne, for example, you can see he has balanced everything very carefully. That is an intuitive thing.

DRS: Yes, I actually think each media, such as photography, painting, film, and writing, and of course music, all these different means of expression—they each have something particular about them, a thing which is pronounced. Otherwise, we would not need all those different media. So, for example photography is primarily about light. It is the art of light. Lighting is everything in photography. That might not be obvious until you try to do photography professionally. But it is true. Painting is the art of color. You might say it is about other things too, but if you look at a Rothko—his paintings sometimes are just big sheets of brilliant color—then you get the idea. You can have a great painting without form. Perhaps that Rothko is the exception; but you might allow it. Drawing is about form, the art of form. Form and sometimes conveying motion. So what happens is painting is usually coupled with draftsmanship. Great painters like Picasso were very skilled draftsman as well. So these people are interacting with the world, responding to it, through representing form. And so all these media, people tend to gravitate to the particular aspect that they are good at, and usually this is aligned to the basic way they interact with the world, to their psychology, you might say.

DRS: Think about Alexander Caulder. When I was a child, I sent a lot of time making mobiles. I used coat hanger wire and paper mâché and things like that. I never had any idea about Caulder, or that he had more or less single-handedly invented kinetic sculpture. But he did. And what we were doing then, in the 1970s as kids, was more or less responding to his art.

DRS: Indeed. He made the silent and still things—previously only statues, stone, bronze—move.

DRS: Absolutely. So what happened with AI Art? This is very interesting stuff. So armed with our understanding of artist psychology and media and different ways of interacting with the world, we can answer your earlier question: why do artists hate AI Art so fiercely?

DRS: That’s part of it, certainly. There is a wonderful meme that goes something like this: “I want technology to do my work, so I can stay home and paint and draw and dance, not do the painting and drawing and dancing.” So that’s quite interesting.

DRS: I do too. Ultimately technology is supposed to help us, not totally fuck us over by supplanting us. But anyway, what happened with AI Art was this: artist types, people who are kinetic—painting is very kinetic, it involves stroking a brush over and over again on a canvas—and it is also emotional. It is supposed to be feeling-driven. Those people are the opposite, by and large, from technologists. From computer scientists. We haven’t talked about computers, but obviously they involve a lot of thinking from their operators, mathematical reasoning, algorithms—a lot of mechanical things—and these seem completely at odds with emotion. Robotic movement is mostly synonymous with cold, mechanical movement—the opposite of the movement and delicate gestures seen in painting, in dancing.

DRS: It absolutely does not. It is completely different. So for example, with AI Art, the way you interact with the model is with a prompt. A prompt is just a written description, it is cold, analytical.

DRS: Sure. But what I mean is, you cannot cry in front of a computer and have it do what you want. You cannot convince it that way. It doesn’t care. To make a beautiful fractal, you will need a mathematical formula. That’s pretty different from what a traditional artist, say a person drawing in a notebook, is doing. There is another aspect to this, which is what I call Opportunism. Opportunism means to take advantage of a situation that emerges, perhaps entirely by chance; and in doing so, to better one’s circumstances, despite principle or perhaps consequences.

DRS: Sure. So suppose you walk down the street and encounter, just by chance, a $100 bill on the ground. Of course, you stoop down and pick it up. Anyone would, right? But then the question arises, has someone lost it? Obviously they have. So an opportunist puts the bill in their pocket and moves on. They will not question it. Whereas another person, someone more focused on, say, principles, will ask, “Who lost this?” Perhaps they will not even take it. I don’t want to paint the opportunist as an amoral bastard.

DRS: Well, opportunism has a scientific meaning, too. For example, certain animals can only eat one kind of food; whales, some of them, can only eat krill. But some animals can eat many different kinds of food, so if they opportunistically come across something, they can take advantage of it, whatever it is. A Tasmanian Devil doesn’t just eat the meat of an animal carcass: it eats the skin, the bones, everything. Whatever it finds. So opportunism is a survival strategy. I think that is a less moralistic way of looking at it. Opportunism can be a kind of mental flexibility.

DRS: Opportunism is actually critical there. Let’s think about a traditional artist, say a draftsman. That person draws a picture using kinetic action (moving the hand); and they judge the drawing a success if it looks close to the model they are drawing from (if it is drawing from life) or they judge it a success if it looks as they wanted it to. There is an element of opportunism here, but it is small. A good draftsman will tell you, there is not much chance in it. It is skill, and practice, that make a good drawing. So, contrast this to someone running an AI Art program. For them, they just type a prompt. Perhaps it can be very complex, or it can be a few words, “a cat.” And then they get a picture of a cat.

DRS: Well, there is. We’ll get to that in a minute. But what is crucial here is that the cat (or whatever)—suppose the picture is a really nice-looking cat. That’s pretty wonderful. It’s not just a copy of a cat pic, like from a database. It’s a new unique image. Now, the prompter did not even do anything specific. In no way did the prompter really do the “work.” Suppose now the prompter asks for “a cat wearing sunglasses at the beach” and the AI dutifully generates a cat wearing sunglasses at the beach. That is pretty remarkable also. But, think about it, there is not much control here. There’s actually nothing about the output from the AI model that is specific to the prompt. Let me say that in a different way: in order to like and accept the AI Art output, the prompter has to be opportunistic enough to accept the output as being cool, or useful, or whatever it was that they were trying to achieve. If it is not, then they can press the button again and get a new image. That process is iterative. Many art forms do contain—I suspect all art contains an iterative aspect. But with AI Art prompts the iterative aspect is an essential part of the process and almost never are you going to get exactly what you hoped for. I’d say never. It can happen that you put in a prompt and you absolutely love what comes out—but you will almost never get exactly what you imagined. There will always be an opportunistic “acceptance” of what comes out.

DRS: It’s not like that. So of course, AI Art process is totally different and does not even use the same approaches to reality that the traditional artist uses.

DRS: Not in the sense that traditional artists mean, no. I don’t think it is. However, it is a kind of art, it is an industrial art. The same kind of question actually comes up with pottery fairly often. If you invest the time—if you pay your dues, as Bob James, my old Ceramics professor used to say—then you can get more and more control over the process until you are statistically likely to get what you expect. That can happen. There is a learned experience process. You can also start to work more from feeling. For example, you can start prompting with a serious purpose, for social, political reasons. You can also explore emotion—do actual art—through the tooling. But it takes time and serious effort.

DRS: Yes, absolutely. Those covers are made with a combination of generative AI Art made on Midjourney, and then extensively processed in a graphics program—I use Pixlr—and then finally used as a layer on a template so things look “of a piece.” The Sunocide albums look pretty good, if I do say so myself.

DRS: It’s fine, you can say it.

DRS: Everything that’s beautiful has some portion that is weird. Someone famous said that.

DRS: No, not much. I use it for book covers and things like that. I put in two years. I was trying to sell tee-shirts with AI Art on them. I didn’t seem many—like three. So that didn’t work. Probably I didn’t know how to sell things. But mainly I misunderstood the blowback from the art world. They claim AI Art is stealing, so there’s no good way to combat that misunderstanding.

DRS: Mostly. With AI Art, visual art, the scrape the internet for training data. Nothing illegal about that.

DRS: The fact is, anyone who has a webpage on the public internet is very keen for it to be spidered—that’s another word for scraped—why? Because they want people to find it. To read it. To pay attention to it. It has to get into the search engine’s database. Anyone who doesn’t want that can do the simple step of adding the no-index META tag to their page. Indexing engines respect that. They will skip your content. But no one does this simple thing. Every artist is desperate to have eyeballs on their content.

DRS: Perhaps. But not every webpage even meets the criteria of being copyrightable. It’s not clearcut. A lot of webpages are advertisements, or cookie cutter copies of one another with minimal changes. At least, that’s the view of the US Copyright Office. Even if the page, for example, an image on the page, is copyright-protected, it’s not clear that training is anything other than fair use.

DRS: Well, science and inquiry are certainly fair use. Training an AI model falls into that category. Even the thing Google search does—putting your webpage data into a search database—which is a commercial activity—no one seems to have a legal issue with that. Why is that not challenged? They are making money off every page they spider, because they run ads and they also sell your data.

DRS: It’s not ideal. Google should certainly be paying every website they spider, some small amount, for the privilege. But it’s a kind of symbiosis (or perhaps, a parasitism). People accept it. They don’t even question.

DRS: I don’t think so. I am assuming we are talking about publicly available pages. But let’s look at it in a different way. Think about a child. Suppose we have a child—let’s say a music student. And that student, we want her to be exposed to all kinds of music and we want to teach the child how to play. Of course she will need examples, she will need to listen to music! Lots of it. And if then, that child listens and learns, and plays their own version of some song, then we will all applaud. Isn’t that true?

DRS: Sure. I think the critics of AI generative art—AI music is one of those, the algorithms are apparently similar to those used to generate images—those critics don’t understand AI is a learning machine. It’s not conscious, in the way we are, but it does actually learn during the training phase. And what it generates actually is new. With LLMs, Large Language Models, those are a little different, but they definitely make new things. Everything that comes out of generative AI Art is “new” in the sense that, pixel for pixel, it’s new. It has not been seen before. So it makes new things.

DRS: Correct. That’s because (they say) a human didn’t make it. Well, does a human “make” a photograph? Of course not. The photographic process uses a camera. That’s a machine. A human made the machine. Or more precisely, a corporation, people in a factory. Maybe even robots in this day and age. The only thing the photographer- the "enduser" - does is press the shutter! At the right moment, in front of the right things, and of course, in the right light. The photographer organizes all that, too. But the essence of the photography process is operating a machine. Indeed, because photographs are usually merely images of things and people, you could argue AI Art is more creative than photography.

DRS: I know. Things are not as simple as I have laid out. To be clear, I think photography is definitely an art. My point is just that if you want to base copyright on “human activity” or “human creativity,” then you have to be fair about it. The US Copyright Office is not.

DRS: That’s correct. However, you have to look at the totality of things. A lot of what happens in photography is in the camera. Think how complex a tool, a machine, that it is. Surely photographers don’t pay thousands for that Hasselblad without reason. So a good deal of the “creativity” is hidden away in the craft of making cameras. It was a craft developed over hundreds of years. You can think of it as a cultural and technological endowment. This is why, sometimes, if a photographer wants to get back to basics, they use a pinhole camera. They want more creativity, total creativity. So they make not just the picture, but even the camera. I’ve never heard of anyone making film, but they could do.

DRS: You see, our world is deeply saturated in something called the division of labor. Specialization is the name of the game. You and I know nothing about AI model design and construction, even those algorithms are very difficult for specialists to understand. You have to not be just a computer scientist, but a special “data scientist” or a “machine learning scientist.” So what this implies is that thousands of man-hours—probably millions—have gone into exploring that domain and that is all very creative. Science is a creative act. Sometimes that is ignored. Anyway, within system “AI Art,” which includes the prompter (a human, usually) and a computer, and some server and the model and the generative engine to run it, and even a business interest to finance it—there is this tool, which took huge investment in human capital to make, and then it is trained on millions of images (or songs or whatever) which is also the creative bootstrap of human capital, human effort. And all that goes into the button press.

DRS: Well, yes.

DRS: Well, yes, marginally. Like many creatives I had a musical period. [Laughter.]

DRS: Before. I had guitars and so on. At one point, I think 1992—I started writing songs. I even went in and recorded a few in a recording studio. Those recordings are all lost, although I have the lyrics from many of them.

DRS: Yes. But more recently, in 2019, I got back into it. The COVID years also were a time of inward-looking—I’m not the only person for whom that was true. So, yes, I was “doing music.” But my idea of music was recording tracks in my room with a single microphone and a Focusrite Audio card. The concept of performance was completely not on the radar. Still, I pushed those recordings up to Spotify, just out of a kind of masochistic desire to expose my creative efforts.

DRS: No, those were pulled down when I discovered what Suno.com could do.

DRS: I should back up and explain that, although I had followed AI Art trends for a long time, and really got into it with the advent of Midjourney, I didn’t think music was even possible. There were some early exciting ideas, but not really a commercial product. So I stopped paying attention.

DRS: The truth is it took Rick Beato (the YouTube music influencer) to push me over the edge. Rick did a video where he said AI Music had no socially redeeming value at all. And he went to this website—suno.com, he said—and showed how it worked. Now, I totally get Rick’s disgust at music created by machines, but, watching the video, I took it as a bit of a challenge. Was it really as bad as he said?

DRS: I have to preface this by saying my musical practice was really about songwriting. I even wrote a book called Lyrics First, which you published.

DRS: I have to rewrite it now. Anyway, my practice is mainly about lyrics. I’m not much of a guitar player, and a terrible singer, and I hate even the concept of performance.

DRS: I know. But when I tried Suno, I found I could input a lyric, and get something pretty good out of it.

DRS: It can. The model has been trained on high-definition, high-fidelity music in a lot of different genres: Rock, Pop, Reggae, Blues, and so on. And it can generate lyrics using an LLM. So it’s a complete prompt-driven solution.

DRS: He got bent out of shape at the idea people were creating fake artists and fake bands, and presenting those as real.

DRS: Oh it is. The issues run deep. In our society, music is generally considered to be an exploration of identity and authenticity. In that context, there’s a long history going back decades of fakers and frauds. There’s a lot of money and fame at stake. So, sure, people sometimes fake it if they can’t make it. I don't want to do that, however. My goals are pretty limited. But the study of "fake" bands and it means to be authentic would make an interesting book.

DRS: That’s a great example. They could hardly play their own songs. The Monkees were a construct; they existed because of the Beatles, who were a real band and actually made and played music, and the desire of TV executives in the US to have a show that would reflect that image, to capitalize on it. At any rate, the dialectic between individual authenticity and the corporate construct, the corporate product, has always been present in music since it started to be possible to make money from recorded music. A more recent example if this “faker” phenomenon was Milli-Vanilli.

DRS: Yes, I know. But I always felt sorry for them, because the producer, Frank Farian, who I suppose was a legitimate music guy, basically created Milli-Vanilli, the entire concept, and the two guys, Rob Pilatus and Fab Morvan, were more or less used as fashion models.

DRS: Yes, they did. But you have to remember, all music videos then and perhaps even now are made using lip synching. It’s not even a question. They are lip synched. The issue with Milli-Vanilli was that the guys didn’t sing the actual songs either. Farian had organized everything. That and their mixed-race status. Because there was a lot of racial bigotry mixed into the press reporting on it.

DRS: So Gorillaz is a really interesting case. They were a “virtual band.” By the year 2000 something like that was possible. They were actually a real band, in the sense that Damon Albarn and Jamie Hewlett had significant musical talent and connected to the emerging virtual reality of inauthenticity and commercialism. So they intentionally wanted to comment on that.

DRS: I think so. Millennial and Gen-Z. The deep cynicism of Grunge is on display, don’t you think? And extensive digital processing. Things were changing rapidly. At any rate, they were a real band, and eventually started playing in front of live audiences as a band. Such is the hunger for performance. Their “virtual” status did not last long and I don’t think it was controversial. Which means it’s possible to to have music that does not necessarily front in the traditional way.

DRS: Sure.

DRS: Obviously I don’t plan to perform many of my own songs. Probably none. I have no desire to do that. But the fact that I can generate music from my lyrics that sounds good, that is worth listening to—that’s worth a lot to me. And if some day a real band wants to play one of my songs—well, my releases are very good demos of what’s possible. So that’s how I think about my musical efforts. They are demos, certainly. But some of them are easily good enough to stand on their own as songs.

DRS: Sure. Let’s talk about disrupting expectations.

DRS: If you listen to the tracks on Sunocide, you will find, I think, five different genres of music. There’s a great contemporary pop song to open with, called Tomorrow Never Knows, followed by a TexMex version of High in the Afternoon. Goodbye Daddy is a Doo Wop song. I don’t think you will find any albums, even collections, like that.

DRS: I think almost never. The reason is that, well, a normal human band can’t do that.

DRS: Probably not. A lot of people seem to have very narrow ideas about what kind of music they like. My son, for example, who is 18 years old and a Freshman in college, he likes a specific kind of Rap music—I don’t even know how to describe the genre, it’s an aggressive kind of contemporary gangster hip hop. It’s total crap to be honest. Anyway, I wrote a hip hop song for him, but he didn’t like it—it wasn’t close enough to his chosen genre. That sort of thing is a little depressing. However, my view is that it’s good for people to be pushed and exposed to different ideas and different art forms.

DRS: Yes, it’s an absurd idea. But I’m old enough so I don’t care.

DRS: White? Yes, I’m white. But to be clear, I don’t really write hip hop songs. I write lyrics for songs, and then I look to see what genre or genres they work well in. Hip Hop is really good for story telling. Now, as regards cultural appropriation, I don’t believe that it is a thing between artists. For example, I like to cook pizza. I’m pretty good at it.

DRS: Exactly. So, is it cultural appropriation for me to make pizza? Are the Italians being made fun of, is their culture being stolen by my very nice pie?

DRS: I don’t think Domino’s or Little Caesars is stealing Italian culture, either, and what they do doesn’t seem to offend Italians. Rather, they are proud to see their culture propagated and loved all around the world. Now, look, I think Domino’s Pizza is crap, but that’s because I’m a snob. Anyway, cultural appropriation implies that culture can be stolen. But human culture belongs to everyone who is human. There is no “culture” that is the specific property of a one group or another.

DRS: I think the identity of an indigenous person, say a Lakota, or Teton Sioux, is not damaged if someone reads The Sacred Pipe, by Black Elk. And in fact, Black Elk intentionally recorded the rites of the Lakota so that knowledge would not be lost. He also raised his children as Christians, because he said they needed to live in the world that was upon them. The upshot is, if Black Elk is not offended by someone like me reading his book, and appreciating his religion, or even practicing it, as far as that goes—then why should anyone else be offended?

DRS: I know. And I am not interested in forcing them to share their culture. However, it is a truism that if we look at the most beautiful and successful human cultures, they are all mixtures and hybridizations. Melting pots. Look at the Romans—they loved and absorbed Ancient Greek culture. The Indians absorbed many cultures over time, such as the combination of Ancient Greek, early Christian, and Buddhist iconography. We think of Buddhism as Chinese or Japanese, but actually, it originated in India and spread elsewhere.

DRS: They are small-minded sometimes, unfortunately. Multi-culturalism is more or less synonymous with any advanced civilized society. It is necessary for people in contact to blend. To share ideals and also to mix. Now, let’s consider the ideals of the White Supremacists. Do they not want to be a “white-only” culture? Do they not seek some kind of cultural “purity” along racial lines? They don’t want to blend. They want to claim Western music and art is their culture and protect it. Look at what the Nazis did, they had a “degenerate’ art show. Well, a lot of that art is the art I prize the most. Like Ernst Kirchner and Paul Klee.

DRS: Anyone who is opposed to multi-culturalism is unfortunately on the wrong side of history. I don’t really care who they are or what they look like. At any rate, we should be polite and respectful. That is a fact. Obviously making fun of people or ridiculing them is wrong. So don't do that! But being polite doesn't mean buying into someone else's views. I have my own views. Aren't they as valid as yours or anyone else's? So yes, no one owns human culture; it is an endowment that belongs to all of us. And that is wonderful! We can use it, improve it, or destroy it through our actions. For example of use, I have a Klezmer version of Sam I Am on Return of Sunocide. That’s the same album that has my From the River to the Sea. Klezmer is a wonderful musical form. I don’t claim to have mastered how to use it yet. But it is wonderful and powerful. I am trying to find more things to do with it.

DRS: Neither. It is pro-human. It talks about the Moral ABCs of Doctor Bronner and human oneness.

DRS: Yeah. The soap guy. [Laughter.] It’s a hip hop song.

DRS: That’s OK. Racism is not going away quickly. The answers to the world’s problems are easy to state, but difficult to implement.

DRS: I have various things in the pipeline that will be coming out soon. Let's talk again about specific songs. There's a lot to say. I hope that anyone interested in seeing what can be achieved with AI Music and good lyrics should search for “Stone Age Time Machine” on whatever streaming media service they use, and check it out.

DRS: Thankyou.