From Unintentional Music to Unintentional Dreams

Sometimes we begin with a clear intention and a fairly precise idea of where we are going. But if we remain attentive to what appears along the way — an unexpected response, an image, a shift in feeling — the process can lead somewhere much richer than the original plan. What first looks like a detour may turn out to be the real direction.

Reading Lane Arye’s Unintentional Music, I kept returning to one idea: the cracked voice, the wrong note, the sound we never planned. These moments are easy to treat as errors. But they can also be signals. Something is already happening through them, and the work begins when we enter their world instead of trying to make them fit into ours. Night dreams may be the most unintentional communication we have — an image, a figure, an atmosphere that arrives outside the plans of our everyday identity.

I wrote an article about following that signal: about what artificial intelligence can and cannot do with a dream, drawn from a year of building a tool for dreamwork. I sent it to Lane, expecting comments and a conversation about where it might be published. Instead, he wrote about the one element in my dream that I had chosen not to follow.

That reply changed the dream for me. It also changed how I thought about the tool I had built. What began as an article about a method became an exchange in which the dream kept unfolding between two people — which is, I have come to think, the truer subject.

I want to keep both movements here: the original thesis as I brought it to Lane, and what happened to it in his hands.

What I thought AI could do

Let me start with the limits, because they matter more than the possibilities.

AI does not perceive double signals. It does not hear the tone of your voice change when you mention the tiger in your dream. It does not notice that your hand makes a small clawing gesture while you describe being chased, or that your breathing shifts when a certain word lands. A human facilitator works precisely with these moments, where the dreaming process leaks into the present. For a language model, all of this is invisible.

AI also cannot give you lived experience. It can describe an energy, sometimes with surprising accuracy, but it cannot sense whether you have actually entered it. It cannot tell the difference between understanding something and being changed by it.

So when the question is whether AI can be a dreamwork therapist, my honest answer is that I do not know how to even approach that task, and I am not sure it should be approached. A narrower question turned out to be much more interesting.

Early on I made a mistake I suspect many people make. I told the model: “You are a professional psychologist. Analyze this dream.” The answer looked impressive and plausible — and it was essentially a performance of analysis. The model reached for a generic script of what psychological interpretation is supposed to sound like.

This forced me to see something obvious but important. When we say “professional psychologist” to a colleague, those two words activate years of training, supervision, practice, personal work, mistakes, and embodied knowledge — most of which never needs to be spelled out. A language model has no such body of practice. For it, “psychologist” means very little until you define what this psychologist notices, in what order, what they avoid, and what kind of relationship to the dream they are trying to hold.

You can tell a human facilitator: “follow the signal.” For an AI, that phrase is too vague. If you want it to do something reliably, you have to make the method explicit — break the work into steps and define what each step is meant to produce. That was humbling, and it also clarified my own work: building a method for a machine forced me to articulate what I actually do with a dream, including the parts I usually do without thinking.

Within those limits, here is what I found AI does well. Give it a dream as raw text and it can identify the main symbols. It can guess which symbol stands furthest from the dreamer’s everyday identity, though it is often better to ask the dreamer directly. It can then guide the dreamer through association questions: What do you think of it? How does it see you? What does it remind you of? What does it want? And if the system has been given a precise method, it can take those answers and formulate the quality that seems to be emerging — in Processwork terms, the X, the unknown figure or quality around which the dream is organized.

Handed back to the dreamer, this formulation can be a real pointer: look here, this seems to be the place where the dream is asking for your attention. The useful part is narrow but real. AI can help locate the charged place. But I believed that everything after the pointer belonged to the dreamer’s body and to the dreamer. To show what I meant, I used one of my own dreams.

The dream

I was in Kyiv and walked into a metro station, one of those old stations with classical, multi-level galleries. I opened a window and, for some reason, wanted to climb down the lattice from one level to the next. It was forbidden. There were many people around. The police came and began checking the cameras. I walked up and said: here I am. We sat in a room together, but it seems they did not find me.

I ran this dream through the structured flow I had built. The system detected the symbols — Kyiv, the metro, the classical galleries, the lattice, the police — and suggested the lattice as the most charged one.

I disagreed and chose the galleries. My energy was there.

Then came the association questions, and my answers surprised me as I typed them:

It is beautiful when there are many levels, when you open a window and there is so much space in front of you. But it is inconvenient. I want to get down and you will not let me.

And then:

This is not about usefulness. It is about grandeur and beauty. Style and monumentality. There must be a lot of air and space.

The spontaneous association that came was a bird with wide-spread wings.

From this, the system formulated an emerging energy of majestic space: airiness, expanse, monumentality. It stood against my habitual identity, which values practicality and convenience above almost everything.

It was accurate. The vector was right. And still, at first, it was entirely intellectual — a correct label I could nod at.

The real shift came only when I managed to sink into that energy and feel the air. There was a moment of realization: the space is already created. The airiness exists in itself. It is not decoration. It is functional, built into the most utilitarian place imaginable — a metro station, where they could have made only a tube, and instead made something like a cathedral.

From inside that feeling, I suddenly saw the same conflict running through several areas of my life, and touched a quiet sadness I carry about not allowing myself that airiness. The sun, quite literally, started playing in new colors that day.

And here is the honest part: I could not bring myself to that point by sitting in front of a screen and reading an answer. The AI had located the charged place with real precision. The next movement happened beyond what the machine could perceive or facilitate. It did not see my breath. It did not know when a word landed. That part, I concluded, still belongs to people — to a facilitator, a friend, or a moment of one’s own embodied attention.

That was where my article ended. Then Lane replied.

Lane dreams further

Lane’s answer did not begin by explaining my dream. He first noticed how its central image was already resonating with his own life, and rather than presenting his associations as neutral or authoritative, he showed me where they came from.

I have been in a very utilitarian / metro-like mode of working on scheduling for the next six months. I’ve taken the month of July to not work with clients or meetings and instead to work on a chapter I’ve been asked to write and then later do a 10 day silent meditation retreat. So the metro is starting to get cathedral-like.

Then, cautiously, he entered the dream.

He wrote that when he thought of Kyiv metro stations, he also thought of how they had been used as bomb shelters during the war. He named this clearly as his association, not mine, and referred to Arnold Mindell’s encouragement to dream further than the dreamer, while always checking the feedback.

Lane connects the dream with the Ukraine support meetings he had been facilitating since shortly after the full-scale invasion. In some of those meetings, the group had gone down into what was happening in a person and in the country, and found something sacred, expansive, even winged. At other times, what was needed was something practical and metro-like: a way to move from being stuck in a trauma reaction toward being calm enough to sleep.

The police in the dream made him think of the edge figures that forbid moving between these levels.

Is it disrespectful of all that is going on in the country and in people’s bodies and hearts to shift levels and find something beautiful in the background?

He described one meeting in which a woman was on a video call when a fighter plane roared overhead and threw her into panic. She followed the terrifying, forbidden energy of the plane. As she imagined spreading jet wings, she found not only the power to destroy but also a power to create. The process eventually contributed to a plan for a volunteer organization that went on to support many Ukrainians.

And then Lane returned to the element I had left behind.

I dream into the lattice. Lattice can be on fences that define borders and edges. And can also be supports for vegetables and flowers to grow. They are at once solid and see-through / airy. In a way they could hint at the non-dual essence that bridges the polarization.

In a second message, he continued:

Did I understand correctly that the lattice was outside the window you opened? It makes me think that the lattice could be both the boundary and link between the inner and the outer, the up and the down, the individual and the world

Every opening came with the meaning handed back to me. Please feel free to toss it. It only has validity if it resonates with you.

What Lane’s response changed

My article had drawn a boundary between machine and human. The machine could locate a charged place. The deeper work, I believed, required a person who could perceive breath, gesture, tone, and bodily feedback.

Lane had none of that. Like the machine, he had only text. And still, his response moved me in a way the accurate formulation had not.

For a while I could feel the difference without being able to name it. Eventually one sentence became clear: Lane did not interpret. He amplified.

He stayed with the image and made it more vivid. A lattice is solid and see-through. It can mark a border, and it can support a plant as it grows. He offered possibilities, but stopped before turning them into a verdict. He also made clear where his associations came from — the bomb shelter, the Ukraine meetings, the metro-like rhythm of his own life — rather than presenting them as neutral facts about my dream. Because he did not claim to know what the dream meant, the act of recognition remained with me.

My system worked differently. It gathered the dreamer’s answers, identified an emerging quality, and moved toward a conclusion. Even an accurate conclusion invites agreement. That is what I had done: I nodded.

Lane’s response invited something else. It made the image larger and stranger. It gave me room to recognize, refuse, or remain uncertain. The mystery deepened instead of being solved.

The AI had pointed to the lattice. I had chosen the galleries, and the work there was real. But the lattice remained unexplored. Lane returned to that abandoned element and stayed with it long enough for something new to appear: a boundary that was also a connection.

The dream enters the relationship

The correspondence did not stop at the dream.

When I wrote back that Lane’s response had become a kind of design specification for a new approach, he was delighted. He wrote:

I’m so excited. Also about our emerging relationship.

Then an edge appeared. A voice in his head said:

Who are you, Lane, to talk about an emerging relationship? … Don’t get ahead of yourself or make any assumptions or have hopes about contact or relationship. Who do you think you are? Don’t forget that you’re American and he’s Ukrainian, that you’ve never even had a conversation outside of email. You shouldn’t even be writing this because then he might feel pressured to be nice to you.

He described feeling tangled and uncomfortable. He paused and breathed. Then he crossed the edge and wrote that he was indeed excited about our emerging relationship, whatever it might become.

Zobaczymy. Das wird sich zeigen. Uvidíme. We’ll see. Побачимо.

The dream was no longer only mine.

The police in the metro forbade going down. His inner police forbade naming a connection. The lattice was no longer only an object outside the station window. It was becoming the boundary and the link between two people who had never spoken outside email.

Lane noticed this himself:

Suddenly I’m not just musing about your dream, it is also inside of me and between us, creating a border and seeing through it.

This was the part no system could have produced by following a better set of instructions. Lane was not only reading the dream. He was allowing the dream to affect him. And because something real was at stake — the possibility of intruding, misreading, or putting pressure on another person — his uncertainty was not a stylistic gesture. It belonged to the relationship.

The human contribution was not simply a better interpretation. It was participation, risk, and mutual change.

What the exchange changed

Before it changed the method, the exchange changed something in how I understood the process itself.

I had begun with a clear frame — an article about the limits of AI in dreamwork — and I thought I knew what the exchange with Lane was for. But the most valuable part appeared outside that frame. This showed me how important it is to hold an intention without becoming closed around it. A goal can give an interaction direction, but it should not decide in advance what the interaction is allowed to become. Something unexpected may appear — an image, a response, a tension, a new connection — and if it is given enough space, it may begin to show its own direction.

That is what happened here. The correspondence was no longer only about commenting on an article or discussing publication. The dream entered the exchange and continued unfolding through it. Neither of us could have planned where it would lead.

I began to understand that the same principle belongs at the center of working with dreams. The task is not only to discover what a dream means. Sometimes the more important task is to stay with it long enough for it to unfold: to give its images space, to let an element become more vivid before deciding what it represents, to allow the dream to reveal connections that were not visible at the beginning.

This changed how I wanted the tool to respond. The most important shift was from convergence toward amplification. Instead of rushing toward the best explanation, the system should help an image become more present. It can say something about the image — a lattice can be both a border and a support — but it should ask, not assert, what that might mean for the dreamer. It should offer several genuinely different possibilities and make their uncertainty clear. It should stop before turning the dream into a verdict or prescribing what the dreamer should become. Most of all, it should return the act of recognition to the person.

This does not turn AI into a human dreamworker. It does not give a machine a body, a relationship, or the ability to be changed by what it encounters. But it may allow the machine to occupy a more useful position: not the one that solves the dream, but the one that gives the dream enough space to continue unfolding.

The lattice

I had begun with a clean division. The machine could point; the human could perceive and unfold. That division is real, and I do not want to collapse it too soon. AI can notice an element the dreamer overlooks. It can hold several possibilities without tiring. It can be available at four in the morning, while the dream is still warm. But it cannot know whether an association is alive for the dreamer. It cannot take responsibility for entering a relationship. It cannot be changed by the dream and then speak from that change. These differences are not going away, and pretending otherwise would help no one.

And yet the image the whole exchange kept circling was a lattice: something solid and see-through at once, a boundary that is also a link. It makes me hold the division differently. Machine and human may look like two poles, and the tension between them is worth keeping open rather than resolving. But the tension is not a wall between two separate worlds. It plays out inside a single field — a shared space in which a dream can be pointed at, and unfolded, and dreamed further, by whoever is present to it. The lattice both separates the levels and lets us see and reach between them. What I want to explore now lives in that space: not which side of the boundary a task falls on, but how the whole process moves across it as one thing.

An AI pointed at a lattice. I disagreed and followed the galleries, and the work I did there was real. Then a person on the other side of an ocean dreamed further, returned to the lattice, and helped something new appear: a boundary that was also a connection. Between the levels. Between inner and outer. Between him and me. And between what a machine can offer a dream and what only people can do with it.

Whether the lattice was the most compelling element of the dream, I still do not know. It may simply have been the most mysterious one. That is a different thing — and giving mystery space is not the same as knowing what it means. It is, perhaps, a Processwork-like attitude toward this strange new instrument itself: to follow what actually happens through it, and to let its unintended shape keep teaching us something about our own.

Afterword

Lane Arye

It has been and continues to be a pleasure, an honor, and a privilege to be in this conversation with you. Thank you for inviting me to have the last word.

I have had an aversion to AI, seeing it as “other” and dangerous. You have opened my mind and my heart in many ways, helping me to see it as part of the field. It has been foolish of me to wish that something that is would not be.

Sitting with your invitation, I found myself meditating on language models that focus exclusively on words. What arose was a sentence from Brook Ziporyn’s translation of Laozi’s Daodejing:

~ Left to themselves, all words fade to silence.