Great post. You are really showing your range as a thinker and writer. here I love studying the rise and fall of different literary and visual styles/aesthetics. I have found Roman Jakobson's theory of the dominant to be quite useful over the years. He defines it as the central focus of a work or literary movement. As a formalist, Jakobson regard the feature as always in the text. As a rhetorical theorist, I tend to define focal points in terms of formal aspects and author-audience processes. Thinking of a dominant in terms of the dynamic interaction between intra- and extra-textual processes gives us greater precision --- and you are very much working in that tradition.
My mentor Jim Phelan defines literary change as a function of 4 interrelated dynamics: 1. extraliterary forces -- the stuff of history, 2. intra/interliterary dynamics -- debates between different schools and between different authors in the same school, 3. the innovation on change --- particularly since modernism, and 4. authorial experiences, preferences, choices, and purposes. To this list, we might need to add automated processes like AI. Or perhaps we can think of them as a function of history. I am not so sure how to deal with the intentionality issue either.
The additive nature of this new aesthetic reminds me of 1950s/1960s chance/probability-based aesthetics where randomized processes drove choice and creation in poetry, music, and cut-and-past novelistic experimentation. Some artists in the 1970s and 1980s enlisted early computers to do some of this randomization, but late postmodernism of the 1990s replaced these completely random surfaces with highly ornate, more baroque, encyclopedia surfaces. Think about the difference between John Barth's Lost in the Fun House vs. David Foster Wallace's Infinite Jest.
When I look at the images in your post, my gut says... AI-assisted Surrealism. Magritte fused with Escher. Magritte's emphasis on the substitution of expected forms with unexpected content. With Escher's orderly regressions embedded inside. In other words, I can clearly see the source material rising to the surface.
This is a really interesting moment in literary/visual history. In the case of unassisted human creation, the process through which Phelan's levels 1 and 2 flow to the surface of work are much more diffuse, but when AI is involved it is much more concrete and transparent. To me, this transparency seriously threatens any real claim to Phelan's level 3, innovation, despite level 4, the intention of the human working with AI. I think until we have AI products that are not simply re-organizing visual inputs in complicated but ultimately explainable way, but offer something close to the spark of creativity that an actual artist experiences after 10000 hours of lived human experience, practice, and cultivation, we will not arrive at a truly AI-inspired innovation in literary/visual aesthetics.
But that is just my two cents. Thanks for giving me the opportunity to think these things through...
Amazing comment. There’s a lot to unpack here but I love the idea that today’s models are better described as distilling extraliterary forces and interliterary dynamics in a way that masquerades as pure innovation.
The idea that the next level of AI generated - or mediated - aesthetics might need to be premised on experience is an interesting one. I’m not quite sure with how we equate the thousands of hours of experience that advanced models see from the largest training runs vis-a-vis human creativity and learning, especially with respect to multimodal models. After all, it’s one thing for a narrow model like Stable Diffusion to create a piece of art versus a much larger, general model that - in the next few years - is likely to take on a more agentic profile.
Whatever the case, though, thanks so much for sharing this. Loads for me to go and think about!
Thank you for writing this article and explaining it with an open mind. I am working as a writer and sometimes feel a chill fear of what AI can achieve, which is much better than what I can do. However, I don't use my fear to judge the technology evolution here. Most of what I was concerned about was easier to see, like when AI suggested something that I can point out coming from other writers, which put users in the position of a potential thief (sometimes users don't know because they didn't read that book/story), and it is unfair to the writer who created the work. But in this case, you tell a different story, which leaves me a lot to think about how we were creative as writers/artists before and now with AI.
Thanks very much! I think I see it both ways: AI will increasingly be used as a creative tool, but it’s not all that clear to me that it’s capable of pushing forwards the boundaries of aesthetics just yet. I imagine that could change in the not too distant future, though.
Great post. You are really showing your range as a thinker and writer. here I love studying the rise and fall of different literary and visual styles/aesthetics. I have found Roman Jakobson's theory of the dominant to be quite useful over the years. He defines it as the central focus of a work or literary movement. As a formalist, Jakobson regard the feature as always in the text. As a rhetorical theorist, I tend to define focal points in terms of formal aspects and author-audience processes. Thinking of a dominant in terms of the dynamic interaction between intra- and extra-textual processes gives us greater precision --- and you are very much working in that tradition.
My mentor Jim Phelan defines literary change as a function of 4 interrelated dynamics: 1. extraliterary forces -- the stuff of history, 2. intra/interliterary dynamics -- debates between different schools and between different authors in the same school, 3. the innovation on change --- particularly since modernism, and 4. authorial experiences, preferences, choices, and purposes. To this list, we might need to add automated processes like AI. Or perhaps we can think of them as a function of history. I am not so sure how to deal with the intentionality issue either.
The additive nature of this new aesthetic reminds me of 1950s/1960s chance/probability-based aesthetics where randomized processes drove choice and creation in poetry, music, and cut-and-past novelistic experimentation. Some artists in the 1970s and 1980s enlisted early computers to do some of this randomization, but late postmodernism of the 1990s replaced these completely random surfaces with highly ornate, more baroque, encyclopedia surfaces. Think about the difference between John Barth's Lost in the Fun House vs. David Foster Wallace's Infinite Jest.
When I look at the images in your post, my gut says... AI-assisted Surrealism. Magritte fused with Escher. Magritte's emphasis on the substitution of expected forms with unexpected content. With Escher's orderly regressions embedded inside. In other words, I can clearly see the source material rising to the surface.
This is a really interesting moment in literary/visual history. In the case of unassisted human creation, the process through which Phelan's levels 1 and 2 flow to the surface of work are much more diffuse, but when AI is involved it is much more concrete and transparent. To me, this transparency seriously threatens any real claim to Phelan's level 3, innovation, despite level 4, the intention of the human working with AI. I think until we have AI products that are not simply re-organizing visual inputs in complicated but ultimately explainable way, but offer something close to the spark of creativity that an actual artist experiences after 10000 hours of lived human experience, practice, and cultivation, we will not arrive at a truly AI-inspired innovation in literary/visual aesthetics.
But that is just my two cents. Thanks for giving me the opportunity to think these things through...
Amazing comment. There’s a lot to unpack here but I love the idea that today’s models are better described as distilling extraliterary forces and interliterary dynamics in a way that masquerades as pure innovation.
The idea that the next level of AI generated - or mediated - aesthetics might need to be premised on experience is an interesting one. I’m not quite sure with how we equate the thousands of hours of experience that advanced models see from the largest training runs vis-a-vis human creativity and learning, especially with respect to multimodal models. After all, it’s one thing for a narrow model like Stable Diffusion to create a piece of art versus a much larger, general model that - in the next few years - is likely to take on a more agentic profile.
Whatever the case, though, thanks so much for sharing this. Loads for me to go and think about!
Thank you for writing this article and explaining it with an open mind. I am working as a writer and sometimes feel a chill fear of what AI can achieve, which is much better than what I can do. However, I don't use my fear to judge the technology evolution here. Most of what I was concerned about was easier to see, like when AI suggested something that I can point out coming from other writers, which put users in the position of a potential thief (sometimes users don't know because they didn't read that book/story), and it is unfair to the writer who created the work. But in this case, you tell a different story, which leaves me a lot to think about how we were creative as writers/artists before and now with AI.
Thanks very much! I think I see it both ways: AI will increasingly be used as a creative tool, but it’s not all that clear to me that it’s capable of pushing forwards the boundaries of aesthetics just yet. I imagine that could change in the not too distant future, though.