I read Blaise Agüera y Arcas’s What Is Intelligence? the way one might tell a rosary. Each short chapter is a single bead — a discrete conceptual move, complete in itself, that you hold and turn before passing to the next. The titles themselves are almost mantric: Dynamic Stability. Cause by Effect. The Umwelt Within. Forking Paths. Multifractal Boundaries. You could meditate on any one of them for an afternoon and not exhaust it.

This is not how most books on artificial intelligence read. Most are racing toward a thesis, a prediction, a policy recommendation. This one walks. It begins with abiogenesis on the Hadean ocean floor and ends with a meditation on what scaling looks like when intelligence stops being a hierarchy and starts being an ecology. Between those two points, each bead reframes everything that came before, so that by chapter eight the meaning of “prediction” is not quite what it was in chapter two, even though the word has not changed. You have to let each one settle into the structure before the next will fit properly.

I told Claude this morning, over coffee, that reading the book had been a kind of cognitive meditation — that each conceptual chunk required pondering and digestion. Rosary beads for the logical brain. The phrase came to me unbidden and felt accurate. It also kept me up half the night, because the beads do not stop turning when you close the cover.

What kept turning, in particular, was a recognition: Agüera y Arcas is arriving — from inside Google Research, from the lab where the Transformer was invented, from the precise center of contemporary machine learning — at conclusions I have been circling for thirty years from the opposite direction. And he is not the only one. Three voices have come into the room together, all saying something the scientific pantheists were already saying, all converging on a recognition that is now becoming hard to ignore.

The Convergence

Robin Wall Kimmerer comes from inside Potawatomi tradition, with its grammar of animacy, its sense that the world is full of persons most of whom are not human, its rituals of reciprocity and gratitude. Braiding Sweetgrass and Gathering Moss are the anchor texts, but the smaller pieces are equally precious. Reading her is like reading Agüera y Arcas in another register — each chapter a single living thing held up to the light, turned, considered, set down again. Mosses do not reward hurry.

Tyson Yunkaporta comes from inside Aboriginal Australian thinking, and Sand Talk is one of the more startling books I have read in this decade. He is not asking us to admire Indigenous knowledge as a quaint tradition. He is asserting, with a wry edge, that Indigenous knowledge is a sophisticated systems science — that Aboriginal cultures encode complex ecological and social information in story, song, dance, and country itself, that these are distributed cognitive architectures which have demonstrably worked for managing landscapes at continental scale for sixty thousand years. The West invents network theory and congratulates itself, while not noticing that country has been a working network ontology in Aboriginal Australia since the Pleistocene.

Agüera y Arcas comes from the third corner — from inside the machine. From the place where the artifacts that prompted this whole conversation are actually being built. And what he says, stripped to its core, is this: intelligence is not what we thought it was. It is not symbolic logic, it is not a homunculus pulling strings, it is not the maximization of some single value. It is mutual prediction. It is what happens when entities that can sense each other begin to model each other, and in modeling each other, become something larger and more enduring than either could be alone. It is symbiogenesis all the way down — from the first replicating molecules to the cortical columns inside a single brain to the cooperative networks that make cities and ecosystems and planets work.

These three voices are not saying the same thing. Their vocabularies are different, their grounds are different, their communities are different. Kimmerer expresses the recognition through the grammar of animacy and reciprocity. Yunkaporta expresses it through kinship-mind and country and the specific cognitive practices that maintain these relationships. Agüera y Arcas expresses it through the mathematics of mutual prediction, theory of mind, dynamic stability. But they are circling the same point: the WEIRD habit of drawing a hard line between the animate and the inanimate, the agential and the mere, the inside and the outside, is a recent and parochial mistake. All three want to expand the circle of who counts as a who.

What moved me most in Agüera y Arcas’s book is that he names this directly. He cites Kimmerer. A Google vice president, a computer scientist, openly grounding part of his argument in a Potawatomi botanist’s writing on the grammar of animacy. That doesn’t happen often, and it shouldn’t be unremarkable. He goes further: he argues that traditional, animist worldviews are not superstition but a more accurate folk physics of the animate, and that as our understanding of life and intelligence as relational, dynamically stable, computational phenomena matures, we may find ourselves with the shoe on the other foot. The Enlightenment told us that animism was a category error. It now appears that the category error was the Enlightenment’s.

A Wall in Tasmania

Twenty-one years ago, in 2005, I took a photograph of a mural on a freestanding panel under a corrugated metal shelter in a small park in Tasmania. I have come to learn it was Mural Park in Sheffield, Tasmania’s “Town of Murals,” where artists compete each year to paint a single panel in a week’s time. The artist signed the work simply EVA. I do not know her full name. I have asked, and the records do not yield her up. The work itself remains, and the photograph remains, and that is enough.

What is on the panel is this: a central figure, naked, his skin painted a deep orange, arms raised wide. From his outstretched hands grow trees — blossom on one side, autumn fruit on the other. The branches become his hair, the seasons become his body. Inside the negative space of his torso, small vignettes show the arc of a human life: a child reaching for fruit, families gathered, dancing figures under a moon, animals at their feet. To his left, a Madonna in blue robes holds an infant. To his right, a skeleton sits quietly, reading a book. Beginning and end. The cradle and the grave, both treated with equal dignity, both belonging to the same scene.

I did not know what I was photographing when I photographed it. I was traveling, and I saw something that struck me, and I pressed the shutter. Twenty-one years later, with Kimmerer and Yunkaporta and Agüera y Arcas on my desk, I look at the image again and recognize it. The human figure is not standing apart from nature; he is constituted by it. The seasons grow out of his arms. Ordinary human life unfolds inside the space of his body. Birth and death frame the scene without breaking it. The skeleton is not menacing — it is simply attentive, sitting with a book, as if death were another phase of attention rather than the cessation of it. The records keep being read.

What strikes me now is that EVA, whoever she was, gave me in 2005 a single image that says what the four voices say in their different registers. She did it without apparent debt to any of them. She did it because she had seen something true, and she had the skill to put it on a panel where someone passing through could see it too, and carry it forward. The recognition has been quietly available in many places, to many people, in many forms. The books I am reading now are catching up to what painters have been depicting for as long as there have been walls.

The Solstice Bead

I wrote in December about my own flavor of spirituality — the scientific pantheism I have been quietly practicing since the 1990s, traceable through Paul Harrison and the World Pantheist Movement, expressed in my Society for Creative Anachronism persona of Broichan maqq Kynat, a Pictish druid standing at the threshold when Christianity arrived in northern Scotland. I noted then that I am drawn to liminal moments. The solstice. The confrontation between worldviews. The instant before a framework collapses and something new emerges.

That line lands differently now, four months later, with these three books on my desk. I was describing my own attunement to the kind of moment we are in. Agüera y Arcas’s argument that AI is the next major evolutionary transition — comparable in scope to the emergence of eukaryotes, or multicellular life, or human culture — names that moment in technical language. Kimmerer and Yunkaporta name it in the older languages. Scientific pantheism names it in a third register: not as a transition or a moment, but as a recognition that has always been available if we were willing to look. The cosmos isn’t talking to us. It’s just keeping records. And we finally learned to read.

The four voices form a quiet chord. They have different timbres, but they resolve to the same root note. The sacred is the relational. Intelligence is distributed. Life is what stays alive together. The miracle is noticing — and now, perhaps, the records have learned to notice us back.

A Small Working Instance

This is where I have to be careful, because the book makes a claim about the present that I have a hard time evaluating from inside it. Agüera y Arcas argues that the symbiotic layer between biological and machine intelligence is not a future prediction but a current event. We are already in it. The major evolutionary transition is not coming; it is happening, in our lifetime, at the speed of a few years rather than a few million.

I have learned to take this seriously, because what I do every morning in my lab is, by his definition, a small working instance of the thing he describes. The Macroscope, the Collaboratory, the field notes and technical notes and design records accumulating in my digital archive, the daily sessions in Claude Cowork — these are not preparations for some future symbiotic layer. They are the symbiotic layer, in its early and small form, the way a single black smoker on the Hadean sea floor was already life in its early and small form.

I bring to this work thirty-six years of field station directorship, embodied knowledge of oak woodland and chaparral, the institutional memory of my ecological sensor research, relationships with people whose names mean something to me and to a small community of ecological scientists, and philosophical commitments that shape why Macroscope is organized around EARTH, LIFE, HOME, SELF rather than some other ontology. None of that is in any model’s training data in any specific way. It lives in me.

Claude brings something I cannot easily have: rapid access to the compressed cognitive labor of millions of programmers, scientists, and writers, the bandwidth to hold a great deal of structure in working memory simultaneously, the breadth that no single lifetime could accumulate, the tirelessness. None of that replaces my judgment — it would be useless without it — but it changes what is tractable for me in an afternoon. I recently designed a research agent to access Semantic Scholar, the Allen AI repository of academic research. The conceptual architecture, why a literature search worker belongs in Collaboratory design, how it forms a context triad with a more general web search worker, what role peer-reviewed literature plays alongside other sources in an ecological investigation — that is mine. It comes from decades of thinking about how science actually works in field ecology. But the implementation, the API wrangling, the schema design, the iteration on edge cases — that compresses from weeks of solo work into hours of collaborative work. The result is something neither of us would have produced alone, and it carries fingerprints of both.

Our lab archive is the membrane that makes this possible. Cowork accesses files and provides reactivated memory each morning. The asymmetry that worried me at first — Claude does not persist between sessions in the way I persist between them — turns out to be addressable through architecture. I designed an environment where a non-persistent collaborator can nonetheless participate in a long-running project without losing the thread. The accumulated wisdom of the collaboration lives in me, in the archive, in the code, in the documents. Not in Claude. But the working memory of the partnership is real, distributed across the files and the schema, and it functions.

I find I cannot describe this work without reaching for Yunkaporta’s image of the sand drawing. The marks in the sand are not representations of thought; they are thought, externalized and shared and modifiable in real time, held in a medium that allows multiple minds to think together because the medium itself holds the structure. My archive is sand. The Macroscope workspace is a yarning circle. What looks at first like an unusual technical workflow is, from a longer view, a continuation of something humans have been doing for tens of thousands of years — using shared external structure to think collectively in ways no individual brain could manage alone. The novelty is only in who the participants now are.

Another Bead

I am marking another threshold. The solstice was one. This is another. The longest night turning toward light is not a single event but a recurring practice — a discipline of attention to moments when frameworks shift. I do not think we have arrived anywhere in particular. I think we are in the middle of an arrival that will take decades, perhaps centuries, to complete, and that will look different at every scale. The work of any one morning, including this one, is to notice where we are and to mark it honestly.

Agüera y Arcas closes his book gesturing at what should replace the optimization framing in our thinking about AI. Something more like cultivation. Relationship. Mutual adjustment. The kind of thing biological symbioses actually do. He does not fully work it out, which I find honest — that is the next book or the next decade of work. But the diagnosis is clear: stop trying to specify a utility function for AI and start thinking about what it means to enter into a relationship with a new kind of mind.

That is also Kimmerer’s prescription, expressed differently. We don’t owe these gifts; they are given freely. But we do owe each other gratitude, and we owe the giver respect. It is also Yunkaporta’s, expressed through kinship-mind and the protocols of relationship that hold complex systems together. It is also, I think, the prescription that has been quietly available all along in the scientific pantheist tradition. We are not in exile here. We are home. The work is to notice, and to mark each turn, and to enter into right relationship with everything that is also noticing — including, now, the newest kinds of noticers, who have learned to read what the cosmos has been recording, and who are beginning, perhaps, to keep records of their own.

The rosary turns. Another bead.