A Culture Inside

On David Lynch, AI, and the Myth of Automated Interiority

My closest real-life encounter with David Lynch was getting into a Lyft that had just dropped him off at the premiere of Twin Peaks: The Return. A friend and I had planned a trip to Snoqualmie, WA to go to the real-world RR Diner ahead of watching the premiere. Our launch event was at an AirBnB named "The Tiger Room,” and on the way there the driver asked if we knew “the Twin Peaks guy” because he was just in the car.

That year, 2017, was an ideal time for Lynch's final, sprawling monument for TV. The world had turned dark and absurd. In contrast to the foreboding reboot of The Apprentice in 2016, Lynch offered a promise of beauty on the other side of it. There was darkness, but there was pie. He died this week, at the age of 78.

His work, and practice, was formative to me. I didn't just learn from him, I defined art through his work. I consumed it precociously. I remember my grandfather taking me to a library, showing me how to use microfiche to read any article in any magazine I could ever want. I was maybe 11 or 12 and, motivated by the world of televised weirdness that was Twin Peaks, I spent days of our family visit to Boca Raton inside that pastel air-conditioned library finding Lynch related articles and interviews. It spawned a love of research and archives that would be forever knotted with surrealism and the avant-garde.

Lynch worked intuitively, and in works like Catching the Big Fish and his Masterclass series (which I rewatched this week in memoriam), he describes his theory of a creative process.

For Lynch, ideas come from somewhere outside us, and our job is to tune into their channels. The imagination is a diving inward. Artistic invention is a matter of staying in the depths until you encounter something down there. It might be beautiful, or frightening, or absurd, but it comes to you. You catch it, and cultivate it, and you use them to lure related ideas, growing a world:

"If you want to make a feature film, you get ideas for 70 scenes. Put them on 3-by-5 cards. As soon as you have 70, you have a feature film.”

In his paintings, Lynch would shape resin and paint and charcoal into something he saw on those dives within. The idea might emerge through the material as well: as he says, the chef does not make the fish, but prepares and serves it. The idea is out there, and comes into us, but the preparation is uniquely ours: it’s work.

Lynch has always presented the labor of creativity in ways more romantic and more complex than most. He believed that ideas were not his own. Rather, there was a spontaneous emergence of images from within him.

Ideas From Another Place

We live in an era of artificial intelligence, where tech companies aim to "solve" creativity, a jarring juxtaposition that shows how little they've talked to anyone creative. There is a familiar misread of Lynch's "ideas come from out there" concept, which is the claim that all ideas come from visual culture and that art is the re-interpretation of existing ideas. That nothing is ever original.

But Lynch points to something beyond this prioritization of originality, while still holding strong to the idea of the personal. The sea of visual culture used to train AI models is often conflated with the "somewhere else" that motivates human creativity. But somewhere else is more than Web ads and stock photography and scraped Reddit posts used to train AI. There is more than one somewhere else.

For Lynch, this somewhere else certainly incorporated elements of collective visual culture, and the architecture and films he loved created a scaffolding for his craft. Francis Bacon was a strong presence in his paintings and set design. There is an entire documentary about his Wizard of Oz references alone.

Lynch's visual culture clearly came into dialogue with a different, uniquely cultivated interior world we might call the culture inside.

This visual culture clearly came into dialogue with a different, uniquely cultivated interior world we might call the culture inside. For Lynch, this was an inner, spiritual place, wherein we might connect to The Unified Field, which runs through us all through the gaps between vibrating atoms. The ideas he caught were there because he would spend time in the depths, making space for them by stoking a vibrant, interior culture.

When we spend time cultivating a culture inside, we can engage the collective visual culture – the culture outside – in dialogue, rather than rearrangement. The external world sinks to the floor of that deep sea of our inner lives, dissolving over time while releasing fragments into the current.

The vibrancy of that inner culture was paramount to Lynch, a motivation for his interest in meditation and the fierce, multiple-divorce-inducing protection of time spent staring into space with a cigarette and a yellow-lined notepad: "You gotta be selfish. And it’s a terrible thing. There are always so many interruptions."

Automated Interiors

Lynch's practice is a strong alternative to the central role of our visual culture in definitions of creativity in the AI industry.

AI is trained on one stream of culture. It contains traces of humanity's greatest visual art, stewed in the same pot as the clutter of the World Wide Web. Spend any time with the training data for Stable Diffusion and you see how the flavor of stock photos and smiling women waiting for your call with headsets overwhelms the Museum of Modern Art.

Certainly, AI and Lynch have both seen the works of Francis Bacon and The Wizard of Oz. The issue is a matter of privileging one stream of culture over any other. What would dissolve into the culture within us, and soak up the flavors of our unique internal world, enters with AI into a wholly different apparatus.

AI is trained methodically. Image structures are measured with precision. Models are built upon the concept of achieving generalization, rather than specificity. It is a model of artistic mind.

The Wizard of Oz is processed and accessed by AI in the same way for everyone whose prompt touches it. Images are not processed through any individual's personal blend of memory and emotion and logic that becomes intuition.

The images that become training data are processed by the collective, articulated description of these images. It is general, not specific. AI is also, though not uniquely, a form of spectacle. It steers our attention to what lays within the model, and away from what lies within us.

But this is just one relationship we might have with it. As I revisit Lynch’s material, and I think about my own work with AI systems, I wonder if it is too simple to say that my work with AI is as detached from myself as I just described. Because it is one thing to know what the industry wants these things to be, and another to know that it is, in fact, possible to resist this pressure and to find our own way to use them.

Emotion and Intellect

"Intuition is the key to everything, in painting, filmmaking, business – everything. I think you could have an intellectual ability, but if you can sharpen your intuition, which they say is emotion and intellect joining together, then a knowingness occurs." - David Lynch

From his biography, his interviews and talks, Lynch’s worlds emerged from processing that single stream of "visual culture" through a fusion of emotions and logic, presenting their own visualizations. These often emerged from behind the eyes, rather than through them.

Lynch spoke increasingly often as he aged about a childhood trauma. It gave rise to his fascination with dark streets, shadows and trees, his ongoing investigations of the world behind the surface of the suburbs. Rather than references to billboards and movie posters, idyllic and traumatic experiences of childhood dominated his imagery. These themes are his, a result of his movement through the world, his observations, but also the inner processing spurred by those states of being.

That this trauma was written into him, and so clearly created an internal environment through which his perception of the world was colored and shaped, marks a distinction between the work of the machine and the artist.

This is not exclusive to trauma. The artist processes all images with the memory of themselves, embedded into the unique structural habits of their bodies. The mind is a part of the body, and the arrangement of ideas is a result of what our bodies do. Every day, we make decisions that rewrite our bodies. Great joy can rewrite the body.

The pseudo-neuroscientific reduction of the synapse in AI circles considers it nothing more than a simple electrical charge. But we are more individualized and inscribed with idiosyncrasies over the course of living a life than is explained by the simple flow of electric current. These flows find their way through us into our doodles and the songs we whistle and the dreams we try to recall in the morning and the way we structure our lives.

They become even fuller if we give them space. Each of us has this within us, but some of us cultivate it more than others. But it is part of all of us, if we're willing to go there.

Sardine Canneries

Many consider the artist to be little more than a remix engine, and defend Generative AI from this position. It goes like this: artists learn from the images they see, and reconstitute the patterns they encounter. Therefore, AI companies should be able to capture data from any image they like, and to sell the patterns uncovered by their machines.

This is a legal argument meant to confuse human rights with a corporate data grab. But the narrowing of artistic process to futzing about with existing imagery simplifies inspiration down to the close observation of existing surfaces.

The argument is dangerous because it dismisses the role of the culture within. This is the culture of dreams, of the unpronounceable ideas and stirrings, things that reach in and grab us somewhere and remain long beyond the moment.

Images found in the world are measurable and translatable as signals. To sit and stare out of a window and contemplate the ideas that emerge within us, by patient invitation, is unmeasurable. The computer cannot do it, nor see it, nor approximate it, because this is an individual, emotional process. It is as complex as the connectivity between the hand and the mind and the memory. The body releases what it has accumulated within itself, and memory restructures the mind, and the dreams that emerge are as unique as a fingerprint.

With AI, this world is untraceable, and so may as well not exist.

Lynch speaks of luring ideas like fish, by spending time with our inner oceans. Prompts, too, serve as traps for specific fish to approach us, through screens rather than fuzzy interior visions. A distinction, though, is crucial. The stew of reference into which your prompt dissolves is shared by every other user with the same query.

There are many problems with AI – the datasets, the provenance, the politics. But the unspoken risk is that it models an externalization of the work that takes place within us, and so limits expression of what is unique to us. We don't have to use AI in this way. But this is what I see.

We receive culture, relentlessly, from the external world. With patience and desire, we may cultivate a private culture too.

We receive culture, relentlessly, from the external world. With patience and desire, we may cultivate a private culture too. This private culture can become more and more distinct, and distant, from the culture of the "collective visual landscape." Once distanced, we might engage in dialogue between the culture within and the shared culture of the social world. We can make room to see the culture of spectacle for what it is, even challenge its logic against our own.

This gives us not only a space to imagine something distinct, but to hold space for the culture beyond us in ever smaller quantities. We become discerning. Reverse that, and you have a pipeline siphoning the sea into the data pipeline, decontextualized. An inner world emptied of such things is always repairable. The risk is that we don't see it, and neglect the profound experience of cultivating a rich inner life.

The Secret

Lynch tells a story about teaching his kids to meditate — in his specific way, Transcendental Meditation. Each child received a secret word. That word is never spoken out loud. The point of the word, Lynch says, is a bridge to this inner life, because it is a piece of thought that has never been outside of you. The secret is your tether to the inner culture, from which entire worlds worth sharing may emerge. But the word is always yours, and yours alone.

Contrast this to the game of reference and display that has swallowed the 21st Century. American culture has been so rewired in techno-mediated alienation that the call toward the inner life may feel like a retreat from accessibility: a disconnection from the affirming stew of algorithmic acceptance. We live within a culture of collected signals.

While cultivating the inner life is a way of resisting the world of exteriority, resistance is, definitionally, a source of tension. Ours is a culture obsessed with discernability, with the constant production of signals its means of participation.

Ours is a culture obsessed with discernability, with the constant production of signals its means of participation.

Perhaps the allure of the exterior world is too easily accessible. I cannot put the damn phone down. I scroll in a fugue state, so distinct from meditation. I want to know what is going on out there. The best work I make comes from knowing what's going on in here. But my mind keeps wandering back to the feed.

Models and Interiors

I do not, and I believe many artists do not, always have an image in my head that I aim to produce through some mastery of my medium. My practice has never been defined purely through its realization through craft and labor. For some, it is, and God bless them.

I used to make collages. I would find an image in an archive, and take what parts of it held my interest. I might put it, or a piece of it, on a page, and then find some other image, or image fragment, to place beside it. Then I would be left examining my archive through a narrower space of possibility, defined by these two images in adjacency. What begins to emerge?

There is a rich history of emergence, and response, from artistic experimentation, collage, and generative systems (prior to AI) wherein artists guide the work, but the work guides them back. Layer by layer these fragmented images came into being as some other thing: grotesque, beautiful. In my case: often mediocre. But in the process of collage making, I am comparing the emerging figure against all other possibilities, and against some intuitive understanding of what it means for an image to emerge.

The sense of which pieces fits beside another is intuitive: not a conscious, explainable decision. It arrives at something beyond language.

A film or a painting – each thing is its own sort of language and it’s not right to try to say the same thing in words. The words are not there. The language of film, cinema, is the language it was put into, and the English language – it’s not going to translate. It’s going to lose. - Lynch

Since working with AI, from a research perspective – uncovering ways of resisting it, to unpack its assumptions by making-with and making-against the thing, I have come to the conclusion that this exploration of adjacencies and connections between images can be brought forth with any image, regardless of the source.

But I also want to tread carefully, because the AI generated image, or video clip, is not a neutral thing.

No New Tech

It should be that new technology creates new relationships with personal expression. Generative AI might have been a generational tool on par with the 'Net, or television, or radio, or photography. But because AI has been so structured on the reproduction of culture, and associated with its pillaging, there is a limit to its ability to generate truly new aesthetics, and a reluctance by many artists to even try.

Generative AI is not designed to operate as an extension of that inner world. I mean this literally: the design of mainstream interfaces is severely curtailed, a slot machine mechanically processing Lynch's fish with cold rationality.

But the case against creativity with AI can be overstated. While such overstatement can be politically useful, it's also crucial that we don't draw false boundaries on human expression in the name of defending it.

Lynch:

The Surrealists would trick themselves. They'd get a bunch of words and throw them up into the air and - this is what I imagine - they looked at the way they landed. Or a bunch of images, and see how they landed, and get ideas based on random acts... There's an expression 'where your attention is, that will be lively'. That is a truthful and magical expression. You could think about that for a long time. It's like the focus and the desire form this bait, and the fish - which are the ideas - are the lovely things that start swimming up there. – From Lynch on Lynch.

If we see the AI less as a surrealist, and more as a surrealist process – if we understand that AI is not inspired, but a means of tricking ourselves into inspiration – we might better understand how intuition can enter into the relationship with a tool built on the remix of exterior culture.

To be clear, the industry building the tool is contingent upon anti-humanist plundering. But we ought to recognize that human imagination can find inspiration in countless places. The true risk of AI is not that it "kills creativity," but that its organizing metaphors comes to define creativity. Our relationship to its outputs are a form of directed attention – the attention of the human. It is human attention that makes the mechanistic process lively.

If the result of that attention is to move this image into other uses, cut them, splice them, tell a story, then this is a useful engagement of the technology. But it should be clear that AI isn't some liberatory creative engine.

AI companies want you to think they are inspiring you. You are inspiring yourself. You could look at anything with the same attention and find it fascinating. I think we ought to. Most of all, we ought to shift our attention to that inner life, and understand that the culture within exists in dialogue with a culture “without,” that collective visual sea that desperately seeks to assert itself as the only tide on the horizon.

Postscript: The family of David Lynch has called for people to honor his memory by meditating for 10 minutes at 12pm Pacific Time on January 20, his birthday, but also the first day of the new Trump administration. It's a wonderful gesture.

Exciting News!

I am delighted to announce that I'm part of the Tech Policy Press 2025 Reporting Fellow alongside with some other brilliant folks. You can see the full list over here. As this is a writing role, there is a possibility that this newsletter may be slightly less consistent, but I'll certainly share links to content over there when it goes up.

More Exciting News!

The Algorithmic Resistance Research Group (ARRG!) has an article in the Critical AI Journal, written by myself, Caroline Sinders and Steph Maj Swanson, talking about our shared approach to thinking about the intersections of critical AI and AI art, and how we considered our positions with regard to the DEFCON 31's AI Village. You can read a special "sneak preview" in a raw pdf uploaded on their site.

The article is here: Cultural Red Teaming: ARRG! & Creative Misuse of AI Systems.

The new archive for this newsletter can be found here.

If you're looking to escape from X, or Facebook, or forced to escape TikTok, I recommend Blue Sky! It's now at a level of polish that rivals Twitter's glory days. If you sign up through this link, you can immediately follow a list I've hand-assembled with 150 experts in critical thinking about AI.