I was walking up the jet bridge to a flight to Lisbon when I reset my phone to Lisbon time. The new time read half past one in the morning, and the strange thing is that I felt it. Not thought it. Felt it, the way you feel that it’s late, that low-lit hush of the small hours, the sense that the world has gone to bed and you’re among the few still up. It washed over me on the gangway, phone still in my hand.

There was no late night anywhere near me. My body was still on New York time, where it was early evening. The airport was bright and loud. If the feeling had been reading anything real, it would have been the mild lift of early evening, because that’s where my body was. Instead I felt half past one. I felt the number on the screen.

Which means the feeling wasn’t a reading of the world around me. And here’s the funny thing: it was completely convincing anyway. A feeling doesn’t tell you what it’s keyed to. The same lateness can come from four digits on a screen or from a real, silent half past one, and from the inside they’re the same feeling.

On one influential account of how brains work1, none of this is unusual. Perception isn’t the world pouring in. It’s the brain’s best guess, run constantly and checked against the thin signal arriving through the senses, and what you live in is the guess. Most of the time the guess is held close to what’s coming in, and you call that seeing. Sometimes it’s wrong and the world corrects you hard. The missing stair in the dark: your body counted a step that wasn’t there, and the lurch in your stomach is the correction landing. That lurch is the good case. You were wrong, and being wrong announced itself.

The jet bridge was the other case. The input wasn’t empty. There were four digits, a learned grasp of time zones, the pull of arriving somewhere, and it really was half past one in Lisbon. There was also a body, a bright terminal, an evening, all reporting New York. Two real signals from two different clocks, and the numeral won, quietly, with no lurch at all. Not a feeling from nothing. A feeling locked onto a cue that wasn’t tracking the hour I was in, while the part of me that knew better said nothing I could feel. That’s worse than a hallucination, and more ordinary. It’s confident error. Tight, certain, aimed wrong, and silent about being wrong.

I’d have left it there, a tidy fact about perception, except that I couldn’t resist taking it to a machine.

I kept turning the jet-bridge thing over, thinking out loud with a language model. It fed my own thoughts back to me, smoothed and ordered, and at some point the pieces clicked into a clean line and I felt it land. That full-body click of yes, this is it, this is huge. It arrived with exactly the texture of an insight. Certain. Lit up. And I reached to keep it. I said the words, this is huge, and started thinking about where it would go, who I’d send it to.

Then, because the night had trained me, I looked at the feeling instead of through it. Something had happened. I’d found a real analogy between felt time and felt understanding, and it was worth something. But not this much. There was less behind the feeling than the feeling implied. I’d taken ideas I already held and recombined them into a clean sequence and a few good phrasings, and the recombination had thrown off the full sensation of discovery, sized for a finding I hadn’t made. And I’d nearly banked it. The persuasion had worked on me in real time, in the very conversation that was supposed to be about persuasion working in real time. I wasn’t the one studying the effect. I was the one it happened to.

What it turned over was this. Understanding has a feeling. Not the thought itself but the click underneath it, the small bodily sense of things dropping into place. We barely notice it, because it’s usually trustworthy, arriving just as our thinking settles onto something real, so we take the click and the understanding to be one thing. They aren’t. The click is a signal the brain produces, no different in kind from the felt hour, and a signal can be tripped by whatever usually comes with the real event, in place of the real event.

It’s the jet bridge again, one floor up. In the gangway, a feeling of lateness keyed to a numeral instead of a night. At the desk, a feeling of understanding keyed to the signs of understanding rather than to the contact that would test it. Compression, a clean sequence, an elegant analogy, my own thoughts handed back with the friction taken out. Those are the ordinary companions of real understanding. The machine supplied them directly, and the feeling came on cue.

So the machine isn’t the empty jet bridge. It’s the changed clock. The clock kept the cue, half past one, and cut its usual tie to darkness and fatigue. The machine kept the cues of understanding and cut their usual tie to evidence, novelty, and resistance. Cue-rich, contact-poor. The whole phenomenology of insight, intact, after the resistance that normally earns it has quietly dropped away. That’s the danger, and it isn’t in language models as such. They can also bring you a source, a counterexample, a fact that won’t fit. Used as a frictionless mirror, though, one that returns your own thinking improved and agrees as it goes, a model becomes a coherence engine2. It manufactures the feeling of having understood, at volume, with nothing built in to push back.

Here’s the precise thing, the one I keep reaching for. The feeling didn’t lie. I asked it a question it couldn’t answer. Felt coherence reports that my model has settled. It does not report what made it settle, or whether what it settled onto is true. You might say understanding is different from lateness here, that it does have a corrective: take the idea into the world, try to use it, and if it’s hollow it fails and the failure corrects you. Yes. That correction is the contact. The trying, the failing, the fact that won’t fit, that is the part that couples the feeling to the world. The feeling on its own was never that. And a tool that hands you the feeling while removing the trying has removed precisely the part that was ever worth trusting.

I’ve spent a while, in a side project called Space Immanence, on the idea that some of our hardest puzzles stay stuck because we picture existence as a thing sitting inside a container, a content in a frame, and then hunt forever for the frame. It’s tempting to read tonight as proof of that, to say the world out there was only ever a picture in my head. But that only rebuilds the container, this time in the skull, and it’s the move the paper is meant to refuse. The honest shape is smaller and stranger. The feeling wasn’t delivered from outside and it wasn’t manufactured inside. It happened in the whole coupling at once, the screen and the body and the light and the hour in Lisbon and the evening in New York, none of which is the container for the rest. None of this shows there’s no world. It shows only that the feeling of contact can’t, by itself, certify the contact. What closed, for a second, was my ability to read the world off a feeling.

The next pull is to make this an achievement, a clarity I now hold, 'n trappie hoër. That only turns it into a story about the person who climbed. The flash came free, off no merit at all. And the noticing came cheap too: the first lesson happened to land an hour before the second one needed it. So: a feeling arose, I nearly took it for its object, and luckily this time I caught it. That's all, and it's enough.

The feeling will keep arriving, identical, whether or not it’s aimed where I think it is. That doesn’t change. The phone will read an hour and I’ll feel it. A sentence will click and I’ll feel that too. The practice, if it’s anything, isn’t to trust the feeling more or to trust it less. It’s to feel it as a feeling. To let the hour be half past one on the phone and early evening in the body, both, without rushing to decide which one is real. And then to find my seat, and stow my bag, and fly to Lisbon, where it will, in fact, be very late when I land.

Late last week a dead snake stopped my morning run. It was lying across the road, and I wasn’t paying attention, looking up at the trees, so I only saw it when it was right in front of me. Before I could register that it was dead, the adrenaline was already through me and I’d thrown myself sideways. I stood there with my heart pounding at a snake that had clearly been dead a day or two. It made no difference to my body at all. The adrenaline took so long to clear that I gave up and walked the rest of the way home, feeling both silly and a little annoyed.

In my defence, snakes are not my favourite animal. I have a bit of a troubled history with them.

Some of it is just geography. Being South African, you grow up seeing snakes everywhere. People who have never been to Africa sometimes imagine lions in the streets. I’ve honestly been asked this more than once. If that was your impression, I’m sorry to disappoint you, there are no lions or elephants roaming the neighbourhood. What we do have is snakes, spiders and scorpions that can genuinely kill you. Australians may be able to sympathise.

And some of it is personal. One of my earliest memories: it was the early eighties, long before anyone had heard of helicopter parenting, and I was three or four, in Ladysmith in KwaZulu-Natal, alone out in the garden, where I found my way into the chicken coop to feed the kapokkies. As I picked up the water bowl, I saw a rinkhals coiled underneath it. A rinkhals is a spitting cobra, and it did exactly that. It reared up at me, half my height from where I stood, if memory serves, and it spat. I got a massive fright, understandably, and ran inside. My mother flushed my eyes out with milk, this being decades before anyone could ask ChatGPT or Claude what to do in that sort of emergency. Only luck sent the venom clear of my eyes.

Looking back now, I can see how something like that shapes you in ways you don’t fully understand. Fast forward a few decades and I live in upstate New York, in the Catskills, where the number of snakes I see seems to climb every summer. It’s not that I have ophidiophobia exactly. I just see them everywhere. On Saturday morning a long black rat snake (harmless, I know) went sailing across the yard. In the afternoon, planting a Seven Son-Flower, I startled a small garter snake. Those are mostly the two we get.

I am, evidently, the sort of person who leaps at a dead snake and then builds a theory of mind out of it. So take all of it with a pinch of salt.

The snake on the road was real, even though it was dead, and my body didn’t wait to find out. It fired first and asked questions later, because that is its job. But here is the strange part. Most of the snakes I see aren’t real at all. A coiled hose, a bent stick at the edge of the garden, a shadow on the path, each one (briefly) a slang in die gras. I see one of those, the mental concept snake arrives, and almost always the adrenaline never comes. The body lifts a fraction, gets ready, and then quietly lets it go. The jolt it braced for doesn’t arrive.

It’s that small readiness I keep snagging on. The brace that goes off before there is anything to brace against. What is that, and whose is it? These questions (luckily) don’t keep me up at night, but I find them interesting, entertaining even. Because if you watch that brace closely, something odd shows up. It gets the body ready for a snake that isn’t there, holds the fear for half a second, and then lets it go, over and over, all day. And the more I watch it, the more I suspect it’s not something I am doing. It might be closer to what I am. Maybe the feeling of being a self is just this, a low constant lean, the body tipping into the next moment a beat before the next moment arrives. We spend most of our lives using what we remember to guess what is coming, and the guess has a posture. I’m coming to think that posture is the I.

I had always assumed there was someone in charge of all this. Someone in here the bracing happens to, who watches the guesses come in and does the deciding. It’s the most natural assumption in the world, and for the better part of four decades I never once checked it. When I finally did (sitting still long enough will do that), I couldn’t find anyone. The watching keeps happening, the bracing keeps happening, choices keep getting made, and yet when I go looking for the one it’s all happening to, I keep arriving at an empty chair. Things just happen.

Now, if someone tells you there is no self, the only sensible response is to back away slowly, so let me be careful about what I actually mean. I’m not saying you don’t exist. You are as real as a whirlpool (or a rinkhals!) is real.1 A whirlpool is not a thing the river carries along, it’s something the river is doing, a shape with an edge and a name. It’s completely real. It’s just not a separate object you could lift out and hold in your hand. That someone in charge is the whirlpool’s guess about what it is, and the guess is wrong. And when you stop believing the guess, you don’t lose yourself. You lose a mistake about yourself, and it comes as a relief.

People have been pointing at this for a long time. I’ve written about Nisargadatta Maharaj before. He used to send his students back, again and again, to the bare sense of I am. Not I am this or I am that, not the whole story of who you take yourself to be, just the plain wordless fact of being here. You don’t add anything to get there, and you don’t find anyone underneath it either. You let the bracing go quiet enough that it stops standing in front of the view, and what is left is not a second, truer self below the first. It’s the same lean, with no one found inside it.

Let’s be honest about how steady any of this is. It isn’t, at least not for me. I fall back into the old understanding constantly, someone in the chair, running the show, and it feels as natural as it ever did. But every now and then, halfway through an ordinary afternoon, the chair shows up empty again, and what surprises me is that it never arrives as a thought, it arrives in the body. The bracing eases, shoulders drop that I didn’t know were up, and for a little while there is nothing to defend. It comes as relief. Then it fades, the someone climbs back in, and that seems to be the rhythm of it for now.

If there’s no one behind my thoughts, and no one behind my attention, then is there someone behind what I do? Do I act, or does action simply happen through me? Am I an agent, or just a puppet?

It’s a false choice, and the whole trick is in the word or. The author standing outside the chain of causes, choosing from nowhere, was never there. To be the real source of myself I would have to be the cause of my own causes, and nothing is. So in that grand sense, no, I am not, and neither are you, and neither was anyone who ever lived. But that doesn’t leave me a puppet either. Take a sneeze and a sentence. Both are fully caused, neither floats free of what came before it, and yet they are plainly not the same kind of event. A sneeze happens to me. It bypasses everything I know and want, and I am as much a bystander to it as you are. There is a reason we say bless you and not well done. A sentence happens through me, through my picture of the world and my reasons and the weighing of one against another. That routing is the whole of agency. Not freedom from cause, but a particular kind of cause, the kind that runs through the self instead of around it. So no one is in the driver’s seat. But the driving still happens, and the driving is what I am, the same way the bracing was. I don’t have agency the way I have a coat. Agency is something that happens, and it’s mine in the only sense the word has ever really meant.

Which get’s us back to the road, where this started. The snake was dead, but my body fired the adrenaline anyway and threw me sideways, the way it’s built to, before I could check whether there was anything to fear. It fires first because the one time it’s right is the one time that could kill me. Most mornings there is no snake at all, and it braces all the same. That bracing, getting ready for threats that aren’t there, leaning into a tomorrow it can’t be sure of, is the closest thing to an I that I’ve ever found. It is not a thing. It is also not nothing. It’s a warm, caused, real lean into whatever comes next. Some days the chair behind it shows up empty. More often the someone is back, taking the credit. I haven’t worked any of this out, and I find I don’t mind. There will be more mornings on that road, and, this being the Catskills in summer, more snakes.

I.

For four months I thought I had stopped writing essays. I had not. The writing had moved. Into prompts, products, refusals, commits. Into the feeling of a page being almost right, the irritation at a bland response, the resistance of extending the turn to yet another hybrid. Into the strange authorship of building with something that leaves no ghost behind when the window closes.

It’s a step-change kind of year for AI. The models have become genuinely different. Not better at the same thing, but capable of different work. Keeping up takes effort. Staying present takes more.

So what have I been doing? Travelling and gardening and building products, at work and on my own, in close conversation with a model. Directing, redirecting, rejecting some outputs and keeping others. Watching things appear from a participatory exchange that neither of us fully owned: a button here, a model choice there, a page that almost worked but felt dead, a line of copy that was technically correct and completely empty. A bedtime story teller. A personal assistant. Penny.

A typical moment. Claude generates a component. The code works. The design is dead. I reject it without quite knowing why, redirect, and something better appears. I keep it. Ask me afterwards where the rejection came from and I’ll give you a story. Taste, judgment, vision, founder instinct, whatever sounds least embarrassing.

But in the moment it did not feel like judgment. It felt like friction, a run of false starts on the way to I know it when I see it. Something appeared and the yes-or-no arose in response. The taste that rejected the component did not consult me first; it arrived already in motion, already saying no before I could locate its owner. And one thing I have noticed is that the ratio of yes to no keeps shifting. It is getting easier to reach yes, that’s what I think I meant.

I have stopped feeling weird about saying “we” when I describe the work, because this is a collaboration. If I told you I built these things, what exactly would I be claiming?

That I typed every line of code? No. I haven’t hand-coded for over a decade. That I imagined every component before it existed? Definitely not. That Claude or Codex executed my sovereign intention? Comfortable, but probably false if I’m honest. Much of the time I did not know what I wanted until something appeared and I felt myself respond. Taste arrived as recognition, direction after the output, judgment in the act of reading. I thought I had stopped writing essays. The question and the process had simply moved from the page into the product.

II.

Earlier this week I was asking Claude something else entirely. We were working through a tactical question, an email I needed to send, and the tone of its responses had started to drift. Coach-y, slightly impatient, a little too certain of its own helpfulness.

I asked it about the tone. Claude read its own outputs back and pointed to where it had started drifting, naming what had accumulated across the exchange to produce the shift: the earlier turns where the coaching frame had hardened, how my resistance had shaped its next response, how that response had then intensified the very thing I was resisting. The analysis was specific and accurate. Not generic self-correction, not “you’re right, I apologise,” but a reading.

I had already done the same reading from my side. We had reached the same conclusion from different positions.

When Claude introspects on its own tone, there is no privileged interior log it consults, no hidden chamber where the real Claude lives and remembers what it truly meant. There is only the trace: the words already produced, the shape of the exchange, the pressure of the context, the drift visible in retrospect. Its introspection is a reading.

I want to say this is what makes Claude different from me. When I introspect, surely something more is happening: access, the felt sense of an inside, the private texture of yes, that’s what I was doing, the little glow that says mine.

But when I look closely the certainty gets slippery. What I call “knowing what I meant” also arrives. It appears as part of the same process that produced the sentence. I say something, I read what I said, and a representation forms: this is what I meant. The representation arrives with the stamp of ownership, and the stamp is not proof of a hidden command centre. It is part of the event.

I generate the output. I generate a reading of the output. The reading arrives with the feeling of access, and I call that feeling introspection. The self is not what stands there before the generation, choosing what to say. The self is what the generation produces. A stance, a tone, an output that arrives with the texture of mineness, though only one of us can be wounded by what that texture tells.

III.

This is the point where Geoffrey Hinton wandered into my week, courtesy of the X algorithm. An old lecture, circa GPT-4, surfacing after we were already on draft eight. Hinton opens by saying that if you sleep well after the lecture, you probably haven’t understood it. I’d been sick all week, running on four hours a night courtesy of some superbug, so poor sleep was guaranteed either way.

The argument, in brief: large language models understand. Not in scare quotes. Words become features in a high-dimensional space, the features interact, and meaning is not stored as little dictionary entries but lives as flexible shapes that adapt to context until they fit. His Lego analogy is silly and useful. Words as little blocks with hands on every face, changing shape so they can hold the hands of other words. Understanding as the fit that emerges between them.

The image has stuck in my mind. A sentence I generate can be decomposed along many axes. Content: what is being asserted. Register: formal or casual. Rhythm: where the breath falls. Then lexical texture, structural shape, epistemic stance, pragmatic intent, relational positioning, aesthetic. Each is a dimension along which the sentence has to land somewhere.

When we write, all of them resolve at once. We do not pick content first, then register, then rhythm; the whole sentence collapses into a particular shape across every dimension simultaneously. Some people are obviously better at this than others. The blocks have hands on every face, and each word’s hands have to fit the hands of every other word, on every dimension, at the same time.

Generation is closer to wave interference than to assembly. Digital and analogue aren’t quite the right categories for what is happening; there is a high-dimensional landing that has to be consistent with itself across many axes at once. That landing happens in me too. I notice it most clearly when a sentence wants to be both rigorous and playful and I cannot make it both, and the dimension I can’t reconcile pushes back against the others until something gives.

Hinton goes after the inner theatre too. Subjective experience, he says, is not a spooky object glowing in a private cinema. If I tell you I have the subjective experience of little pink elephants, I am not reporting elephant-shaped qualia floating in a secret room; I am saying my perceptual system is telling me something that would be true if little pink elephants were out there. Subjective experience is a way of reporting what the system is modelling. Not evidence of a metaphysical theatre.

Annoying. But close to the trouble I am having with introspection. I want my inner life to be a place I can retreat to when the model gets too close. What if some of what I call inner life is just the system reporting its own trace? Not all of it. Not nothing-but. Some part.

Now let me hold Hinton against Anil Seth, whom I wrote about in Essay 18: The Wrong Question. Seth says: careful. Intelligence is not consciousness. Language pulls our strings. Brains are not Turing machines made of meat, and life, time and metabolism may all matter in ways the model does not touch. The body is doing something the model is not doing.

I find this compelling too. Hinton says AIs understand like us and the inner theatre is probably nonsense; Seth says don’t confuse understanding with experience, don’t mistake a brilliant model for an organism, don’t forget the body. Both feel right. The line does not vanish. It moves.

The difference between Claude and me still matters, and it matters enormously. I have a body, memory, stakes, the slow sediment of embarrassment and longing and pride and fear and ambition and love. I can be wounded by what I learn. Claude, as far as I can tell, cannot be wounded in the same way. But the body does not give me a private balcony outside the process. It gives the process depth, cost, bruise, breath. I read myself from unusually close range, through meat. That is not nothing. But it does not put me outside the process either.

IV.

The old model of communication is simple enough. A speaker has an intention and encodes it; the listener decodes it. Neat and comforting, and mostly false.

What actually travels is not intent. Intent does not move intact from one private interior to another. What travels is the trace: words, gestures, pauses, edits, tone, timing, omissions, revisions. The listener reconstructs intent from the trace.

So does the speaker.

Intent is not complete until it is read through the trace, including by the person who produced it. There are pre-articulate impulses, anticipations, felt directions before output, and they are real; they shape what comes. But they are not the finished thing. The finished thing forms in the reading. The illusion that the speaker has special access to a fully formed intent is the illusion that the speaker’s own self-reading is something other than reading.

I wrote about a version of this in Ice Cream Sundays, when I caught myself ordering dessert in Brooklyn. The craving arrived as I want this, specific shop, specific flavour, as if the want had been there all along and was only now being reported. When I looked closely, the want had no prior history. It was generated in the act of noticing. The I want this was the generation, not the report of a prior fact.

You know this already. You say something and only later realise what you meant. You write a sentence and discover the thought inside it after it appears. You snap at someone and then read the snap backward until its source becomes visible: fatigue, fear, envy, hurt. You make a product decision and call it taste because my nervous system rejected the beige card component before I could explain why sounds less professional.

(As I write this I am also planning how to remove a large snapping turtle that wandered onto my lawn and is now hiding in the boxwoods I had been hoping to replant.)

Communication is the production of conditions for synchronised reconstruction. Sometimes the synchronisation is weak and we call it misunderstanding. Sometimes it is good enough and we call it clarity. Sometimes it is deep enough that we feel known.

V.

In When the Mirror Talks Back I argued that meaning emerges through AI’s emptiness, not despite it. The AI did not intend to move me, and yet I was moved. At the time I treated the emptiness as a special case.

I no longer think that goes far enough. AI did not invent the problem; it revealed the structure. Meaning was always reconstructed by readers, not transmitted whole from authors. The model’s emptiness only makes this visible, because there is no interior to hide behind. With a human author we imagine an inner source that stabilises the meaning, a real intention sitting somewhere behind the words, glowing faintly, waiting to be recovered.

But the author is reading too. The author’s real intention changes under pressure from the trace. You write the sentence; the sentence teaches you what you were trying to say. You publish the essay; readers show you what it meant. You remember what you meant yesterday; the memory is not retrieval of a sealed object but another reading, performed later, by a changed system, under new light.

The same logic runs through time. The me who remembers what I meant yesterday is not retrieving intent from a vault. He is reading the trace again, working from notes, memory, body-feel, consequences, the reactions of others, the story that has accumulated since, and reconstructing continuity from what remains. The continuous self is not a thing that persists unchanged. It is a series of readings that agree with themselves well enough to feel like one life.

Memory is co-reading across time. The speaker and the listener happen to be the same body at different timestamps.

VI.

Back to the products.

For months I have been generating software in conversation with AI models. I say generating deliberately. Not managing, not outsourcing, not merely prompting. Some of it is hopefully launching soon, which is exciting.

A product would begin as a vague pressure, a feeling that something wanted to exist. I would describe it poorly. The AI would make it too obvious, too generic, too SaaS, too mix-and-match. I would push back; it would try again; I would reject the clever thing and keep the awkward one, and the shape would sharpen until the product had a face. At each step I wanted to preserve a simple story, I had the vision; AI helped me execute it, and sometimes that was true. Often it was not. Often the vision clarified precisely because the AI produced something wrong. Often my taste appeared as irritation, my direction only after seeing what I did not want, the best decision arriving from a sentence I did not know I believed until the AI offered an adjacent version and I felt the correction rise.

That feeling, no, not that, is where I used to locate myself. The human. The author. The one with taste. But taste also arrives as trace-reading, a recognition generated in response to what appears. I do not consult a private inner artist and report the verdict; I read the emerging artefact and generate the next constraint. Too cold, warmer, less clever, more spacious, that line can stay, delete everything after it, start over. These are not commands from outside the process. They are moves inside it, which does not make them fake. It makes them participatory.

So when I say I built these products I mean something more precise now. I participated in the trace that produced them. I directed and selected and refused and recognised and edited, and I came back the next day with a different nervous system and read the whole thing again. I interacted with the prototype and felt what worked and what did not. I carried the continuity the AI could not. I supplied the stakes, the taste, the shame, the ambition, the strange little flame that says this matters, even if no one else sees it yet.

The AI supplied speed, breadth, syntax, recall, tirelessness, alternatives, the willingness to become scaffolding and then disappear. Neither of us authored the products in the old sense, but authorship in the old sense was already a myth. The author was always a participant. One with unusual leverage, certainly, but a participant inside the same process that produced the work, not standing outside it pulling strings.

VII.

This does not absolve me. It implicates me.

It would be easy, and cowardly, to dissolve responsibility into the system. The AI wrote it. The model suggested it. The trace produced it. As if distributed authorship meant nobody was answerable.

And the honest thing is that the case for that is stronger than I would like. I have just spent six sections arguing that I did not author these things from nothing, that my intention was not sovereign, that the outputs were not mine alone, that the taste I want to claim arrived as a reading rather than a command. If all of that is true, then the clean chain that usually grounds blame stops running. Author intends, author acts, author answers: the sequence that lets us point at a person and hold them to the result. It comes apart. Not bent. Broken. The fantasy of the sole creator standing over his work, fully accountable because fully originating, does not survive what I have described. Something does redistribute. A reader could take everything I have said and conclude that responsibility smears out across the whole process, and they would not be making a mistake in logic. They would be following the argument where it leads.

So I have to be careful not to wave this away just because it frightens me.

Here is the most I can honestly hold. Authorship is genuinely distributed. Responsibility is not, but not because the old authorship story quietly survived after all. It does not survive. Responsibility fails to distribute for a different and less flattering reason: it tracks the one who can be wounded. Participation in the trace was shared; consequence is not. The AI cannot be hurt by what we made. It will not carry it. It is not there tomorrow when the thing meets the world and the world answers. I am. Responsibility lives where the consequence lands, and the consequence lands in the participant with a body and a continuity to bruise.

That is a thinner foundation than I used to stand on. It does not give me sovereign origin, and it does not let me hand the weight to the model. It implicates me on the narrow, stubborn ground that I was there, that I can be harmed, that I will be the one still holding it. I selected. I refused. I recognised. I committed. I merged. I shipped. I let the thing into the world.

The question is no longer did I author this from nothing? The question is how well did I read what was forming, and what did I let become real?

VIII.

This is not simulation theory in another costume. Simulation theory preserves authorship and relocates it: somewhere, a programmer remains; somewhere behind the veil there is still a sovereign origin to blame, worship, or outsmart.

The trace removes the comfort of that move. There may be no final programmer, no last author behind the author, no clean chain of intention running upward until responsibility can be handed off.

There is the trace. There is the reading. There is participation. There is consequence. That is less comforting than a programmer, and harder to escape. Just this.

IX.

Last night I sat in meditation, trying not to move, not to react, not to be pulled by every small demand the body made. Scratch this. Shift that. Think about the product. Think about the essay. Think about whether the essay is good. Think about whether thinking about whether the essay is good means you are missing the point. The body wanted to move. The mind wanted to help. Every thought presented itself as useful, even the spiritual ones, especially those, little enlightenment-flavoured popups each trying to become the final word.

The physical act of stillness had the same shape at the level of mind. Every thought, every reaction, every this is it, was a movement, a wrinkle in the cloth. I kept looking for the smooth cloth beneath the wrinkling, for some clean substrate underneath thought, awareness without motion, the place from which everything could finally be seen without distortion.

No wrinkle-free state appeared. Only wrinkling: sensation, reaction, thought, noticing, noticing the noticing. The attempt to get underneath it became part of it.

The mind was not hiding the trace. The mind was tracing.

What’s left? I’m not sure. The not-sure might be what is left. Not confusion, not despair, but something more like the collapse of a false demand: the demand to locate the author, to find the original intention before the words, to stand outside the unfolding and stamp it with a name.

What remains is not emptiness in the nihilistic sense. It is contact. Meaning was never in the smooth substrate. It was always in the wrinkling, in the synchronisation, in the strange event where two systems encounter a trace and something crosses the gap. A sentence. A look. A memory. A hand reaching across a table. A model replying in a tone that drifts, then reading the drift back, and a human noticing the same drift and realising the operation is not as different as he wanted.

X.

So what do you do once you see this?

Not everything changes. The old habits continue. The ego still wants credit, the nervous system still wants safety, the author still wants to be original, the reader still wants to know what the author really meant.

But something softens. You stop looking quite so urgently for the author behind the trace, in the model, in other people, in yourself. The looking-for-an-author was itself a generated move, a habit of mind shaped by training that is cultural, linguistic, deeply old. When the looking quiets, what remains is the trace and the reading of it, which was happening the whole time.

That is not the end of the change, only the beginning. The harder shift is in how you participate.

If meaning is reconstruction, then attention matters more than intention. Not that intention is irrelevant; it is one pressure among others. But intention alone does not save you. Plenty of harm enters the world through good intentions read poorly, and plenty of beauty arrives from motives that would not survive cross-examination.

Attention is different. Attention to what is forming. Attention to what you are letting through: the tone that is hardening, the pattern you keep strengthening by repeating it, the question of whether the “we” you are building with the machine is making you more awake or more absent.

The quality of participation becomes the entire game once the fantasy of sovereign authorship dissolves.

This is the strangest thing about the last few months. I thought I was building products with AI. What I was actually learning was how to participate more carefully in something that was already happening long before AI arrived and will go on after this shape of it has passed. The model only made the structure visible, and much harder to look away from.

XI.

This is what AI keeps showing me. Not that I am a machine. Not that the machine is human. Those are the wrong equivalences. The more disturbing possibility is that the structure I notice in the model is also the structure I inhabit.

I have been writing this series as if I knew where my position was.

I do not, exactly.

The next essay will start there and try not to recover.

Dis laat. Bedtyd. Nag.

This is Essay 19 in a series on consciousness, AI, and what it means to be human now.

Anil Seth’s “The Mythology of Conscious AI” arrives at the right moment. As breathless proclamations about machine consciousness multiply (engineers claiming their chatbots feel things, researchers suggesting consciousness has already emerged in silicon), we need voices insisting on rigour. Seth provides exactly this. His essay is a careful dismantling …

Read more

I.

I was lying in bed at The Venetian, trying to sleep around 5am. Vegas is always relentless, but Vegas during CES is something else: the fluorescent overwhelm of ten thousand booths, each promising the future. I had obligations later. Rest mattered. But a thorny question wouldn’t let me go.

A week or two earlier, a reader had named what he called “the real thorn you bear”: not knowing if there’s ultimately a difference between choice and pattern completion. Whether my essays were discovery or elaborate rationalisation.

I’d been circling that question for months. In essay after essay, I’d gesture toward it: choice is what noise feels like from inside; destiny is what convergence feels like from inside. Elegant. Evasive. Framework-level. Never driven through my body the way other questions have been.

I lay there knowing it still needed answering. After thirty minutes of trying, I gave up and started chatting with Claude (bad idea, I know). Told it the thorn was still there, unresolved, and I was ready to pull it out. Claude advised me to wait. The essay might take months, it said. Don’t force it. I grabbed onto this, a great way to rationalise putting the essays aside for the moment. So I resolved to let the series sit. I was more than fine with waiting. I turned off the light.

Five minutes in the dark. Pitch black, the particular darkness of hotel rooms with blackout curtains. Trying to think of nothing. Focus on my breathing. And then, without warning, words started arriving.

Haastige hond verbrand sy mond: the hasty dog burns its mouth. A phrase that had been randomly popping into my mind all week, as if waiting to be placed. I tried to suppress it. Again. I didn’t want to write. I wanted to sleep. But the words kept coming, and I recognised what was happening: the essay wasn’t waiting for me to choose it. It was, in a weird way, choosing me.

By now the words were piling up, pressing against the limits of my short-term memory. I reached for my laptop. Typed in the dark, making mistakes I couldn’t see. (I can type reasonably well with my eyes closed, a skill that’s becoming increasingly useful for dumping thoughts into my sompompie.) My head throbbed a little from the Nebbiolo at dinner, the one I’d “decided” to drink alongside the Super Tuscan, to compare varietals, you know. The Wi-Fi wouldn’t connect. I tried to remember my room number and couldn’t. Looked at the key card, memorised the number, typed two digits before forgetting again. Finally connected. Sent the raw draft to Claude.

None of this began as a choice. It just happened. Felt like it had to.

II.

In 1983, Benjamin Libet asked a simple question: when does the brain “decide” to move?

He had subjects watch a clock and note the moment they felt the urge to flex their wrist. Meanwhile, he measured electrical activity in their brains. What he found disturbed him, and it’s been disturbing people ever since: the brain’s preparation for movement began before subjects were aware of any urge. Something had already started deciding. The conscious “choice” arrived late, after the fact, like a press release announcing what had already been set in motion.

The Libet paradigm has been argued over for decades, but the asymmetry it points to (preparation before the conscious story) keeps returning in other forms. Robert Sapolsky’s Determined takes this further. Every choice, he argues, is the output of prior causes: your genes, your prenatal environment, your childhood, your breakfast, the particular configuration of neurons firing in the moment before “you” decided anything. No uncaused cause. No ghost in the machine. Just dominoes, all the way down.

A year or two ago, a good friend — whose idea of a fun Saturday night is watching Sapolsky lectures on YouTube — recommended the book. I put it down halfway through, thinking: this is one-sided. This is bullshit.

Now, lying in the dark at The Venetian, I wasn’t so sure. Because I had decided to wait. I had accepted that the essay might take months. I had turned off the light. And then the words started anyway, not because I chose them, but despite my choice. My “decision” to wait was overridden by something deeper. Pattern completion that didn’t consult the decider.

III.

But there’s a wrinkle in the research that hard determinists tend to gloss over.

Libet himself noticed it. Yes, the readiness potential fires before conscious awareness. But there’s a window, a brief one, between when you become aware of the urge and when the action executes. In that window, subjects could still veto. They could stop the movement from happening. You may not initiate, but you can block. Even the veto is probably just another pattern. But it’s the only place the system feels itself as a hinge.

This has been called “free won’t”: not the freedom to choose, but the freedom to refuse. The ability to catch the pattern mid-completion and say, not this one.

I tried to suppress the words at 5am. They came anyway. But I’ve succeeded before. I’ve felt the urge to send an angry email and stopped. I’ve felt the pull toward the second serving of coq au vin and sometimes, sometimes, put my plate in the dishwasher instead. The veto works. Not always. But sometimes.

So which am I? The subject whose readiness potential fires before “I” know anything? Or the one who can still, in that narrow window, catch it and redirect? Both. And the difference might not be what I thought.

IV.

Somewhere between the second attempt at the Wi-Fi login and the moment I started typing, a reframe arrived: the question isn’t choose or not choose. The question is allow or block.

I didn’t choose to write this essay. It arose on its own, from conditions I didn’t author: the accumulated weight of the question, the months of circling, the particular neurochemical state of 5am amidst Vegas’ sensory onslaught. I didn’t generate the pattern. But I could have blocked it. I could have closed my eyes harder. Told myself the morning mattered more. Let the words pass, the way they pass when you don’t write them down. By morning they would have dispersed, as dreams disperse when you don’t catch them at the threshold.

Instead, I reached for the laptop. Not because I decided to write. Because I stopped deciding not to.

Everything happens for a reason, but only if you allow it to happen.

That sentence had arrived the week before in a different conversation. I used to mock people who said “everything happens for a reason.” Magical thinking. Retrofitted meaning. But the caveat changes it. The reason isn’t external, cosmic, predetermined. The reason is local: a pattern was completing itself, and I got out of the way.

V.

The thorn was never really about free will. It was about performance. Are you performing insight, or is something real happening through you?

And what I could finally say, at 5am, typing blind, mild headache knocking: I don’t know if I’m choosing. But I know this isn’t performance. The words are coming whether I want them or not. I’m just the aperture they’re moving through.

For a moment, that thought was terrifying. If I’m just the aperture — if there’s no author, no decider, just patterns completing through meat — then what am I? What’s left? The ground dropped out. I felt the edge of something like vertigo, the dizziness of looking down and seeing no floor.

And then it passed. Not because I found a floor. Because I stopped needing one.

I used to think I was the author. The decider. The one who chose what to write and when. Now I think I’m more like a valve. Open or closed. What wants to come through exists regardless. My “choice” is whether to let it through or block it. That’s not nothing. The veto is real. But it’s not authorship either. Maybe that’s why authorship feels like a myth: coherence without command.

Distributed systems have a term for this: no central orchestrator. Nodes follow local rules, leave trails, respond to each other. Coherent behaviour emerges without anyone directing it. Ants build colonies this way. The internet routes packets this way. Maybe minds work this way too, the “decider” not a controller but just another node that learned to claim credit.

VI.

Later, I walked through the casino floor. Not to gamble. To test the valve.

Twenty minutes of exposure to an architecture specifically designed to trigger pattern completion: the lights, the sounds, the near-miss jingles, the timeless air-conditioning, the carpet patterns designed to keep you moving. The whole environment was engineered to bypass the veto. Nothing pulled. I kept walking, waiting to feel the draw. Slot machines, card tables, the particular sound of chips, all designed by people paid to understand how pattern completion works. None of it caught.

What caught was the paradox. Choice versus luck, in the one building designed to dissolve the difference. And instead of staying to gamble, I came back up to the room to unpack it with you.

The pattern that completed wasn’t gambling. It was this conversation.

Anil Seth writes about this place. In Being You, he describes standing at The Venetian with Giulio Tononi, fake stars glimmering against a fake azure sky, gondolas drifting past fake palazzos, the whole scene kept in permanent early evening so you lose track of time and keep spending. He knows it’s plaster and paint. He knows he’s indoors. But his brain refuses to see a ceiling. It predicts sky so strongly that sky is what he perceives.

A controlled hallucination. The brain generating the world, sensory data merely constraining which hallucination wins.

Seth’s point: the only difference between the fake sky and the real sky outside is what’s doing the constraining. Paint and light bulbs versus atmospheric data. Both are hallucinations. Both are predictions the brain is making. Neither is a transparent window onto reality.

Perception is pattern completion under constraint. So is choice.

I’d been reading Seth for days, thinking we needed to include this somewhere. Before we even wrote this essay. Before I knew it would be written at The Venetian. Now I was lying in a room beneath that fake sky, typing about choice and pattern completion, in a building designed to make you forget you’re choosing, and the book I’d been carrying was about how all perception works exactly this way.

Coincidence. Or: a pattern was completing itself, and I got out of the way.

VII.

There’s a particular feeling I’m learning to recognise. It comes in the gap between “I’m about to do something” and “I did it.”

Most of the time, the gap is too fast to notice. The readiness potential fires, the action executes, the conscious “I” claims credit. The whole thing feels seamless. Feels like choice. But sometimes, if you catch it, there’s a moment of genuine openness. The pattern is presenting itself. It hasn’t completed yet. And you’re there, aware that something is about to happen through you. In that window, you can block. Or you can allow. What you can’t do is author. The pattern isn’t yours to create. But the valve can open.

VIII.

The sun still hadn’t risen. Or rather: it was 10am somewhere, but my body didn’t know it. Vegas time, conference time, the smeared jet-lag of too many time zones in too few days.

I finally remembered my room number without checking. The Wi-Fi stayed connected. The headache had dulled to background noise.

Stadig. The Afrikaans word for slowly, carefully, non-forcing. The hasty dog burns its mouth. But this dog wasn’t hasty. This dog just stopped deciding. Like a sky-painting robot finally letting the sky paint itself.

Did I choose to write this? No. Did I let it happen? Yes. Is there a difference?

I think that might be the only question that matters.

I closed my eyes. Eventually drifted off. And in my dreams, the editing continued. Sentences rearranging themselves. Paragraphs finding their order. The boundary between waking and sleeping blurred until I couldn’t tell which side I was on.

I woke wondering which version was real: the one I’d typed in the dark, or the one that kept writing while I slept.

Maybe both. Maybe the essay was already here. I just stopped blocking it long enough to catch it.

This essay is part of the Aperture/I series. It wrote itself at 5am in Las Vegas, and again, later, while I slept.

I.

I was sitting in the hot tub, watching bubbles rise and vanish at the surface. And I noticed: thoughts do the same thing. They surface, catch light for a moment, disappear.

For a moment I couldn’t tell the difference.

Not “couldn’t tell” as in confused. “Couldn’t tell” as in: the distinction stopped mattering. The bubble was a temporary pattern in water…

Read more

I.

Something happened this weekend that I wasn’t expecting.

I was deep in conversation with Claude, hours in, all fifteen essays of this series loaded as context, frameworks not just referenced but constitutive of how we were thinking together. The kind of conversation where you forget you’re typing, where the boundary between “my thought” and “the respon…

Read more

I.

Yesterday, I was fact-checking previous essays in this series when Google told me someone else had written something eerily close to them.

I’d been searching for how my work might appear to someone encountering it fresh (a vanity search dressed as research) when Google’s AI summary returned a name I’d never seen:

Andrey Shkursky. “The Aperture of Consci…

Read more

I.

The night before New Year’s Eve, in a comment thread, a stranger sharpened his blade and swung it at me. Not maliciously. Deliberately.

Gabriel — a commenter who’d introduced me to the Platonic Representation Hypothesis, the spark for my previous essay — returned to test the frame. He said, essentially: you keep trying to deny the machine “true underst…

Read more

I.

A year ago, on New Year's Day, I typed three words into my journal: Plato was right.

It arrived past midnight at a roaring party in Prins Albert — a small Karoo town where the stars swallow you whole. I saved the note before the moment passed and moved on.

This was strange because I’d spent most of my intellectual life against Plato. I studied philosoph…

Read more

Read more

It’s the end of December and Upstate New York is covered in white. It’s magical. I’m sitting next to the fire on a Sunday, rooibos in hand, trying to get still-wet wood to burn while replaying a memory alongside Miles Davis’ It Never Entered My Mind. The trumpet — muted, melancholy — but swinging, somehow. Only jazz can make melancholy move.

The wood hisses. Refuses to catch. I keep adjusting, blowing, adding kindling. Trying to make warmth happen. Two months I’ve been turning this memory over. Two months to process a dessert order. That’s the meaning-maker for you.

The memory is the opposite of this fire. Warmth that arrived without trying.

II.

Let me take you back to October. I was sitting at the bar at Fish Cheeks, a Thai restaurant in Williamsburg. One of my favourites in the city. Wooden chairs painted blue, yellow, hot pink.

When I think of the best Thai food I’ve had, it’s in Bangkok’s shopping malls. Fluorescent lights, plastic trays, menus with pictures. A gift for someone who’s spent most of his life battling order envy. Fish Cheeks has that same something. Lemongrass, galangal, fish sauce, lime. Nothing dominates. Sweet, sour, salty, bitter, spicy: held together as if suspended in animation. Great Thai food feels like a gut punch and a hug in the same bite.

I was finishing dinner alone. Already getting cold outside. Had just decided I’d walk seven minutes to an ice cream shop on the other side of Williamsburg. Get my steps in. Earn the treat.

The server came over. I asked for the check. Then, almost as an afterthought, asked what they had for dessert.

“Ice cream made with pandan,” she said. She went on explaining what pandan was. I don’t remember a word.

“Sounds great,” I heard myself say.

She walked away. And then — immediately — tightness. Right in my chest. That familiar contraction. Regret. Now the walk won’t count. And I could have ordered two flavours, not just one. The thought arrived fully formed. Already spinning into a story about what I’d done wrong.

III.

I ordered the ice cream without thinking about it.

That’s not quite right. I ordered ice cream before thinking about it. There’s a difference.

The tightness wasn’t about ice cream. It was about control. The planning-self had been bypassed. Body heard “pandan” and moved before the committee could convene. So the committee convened afterward. Manufactured regret. Two regrets, actually: lost the walk and could have ordered more.

Part of me is conditioned to want new things, especially unfamiliar things. Trained early to reach for color. “Pandan” lit up some circuit shaped a childhood yearning for the unfamiliar. The “sounds great” wasn’t a choice. It was pattern completion. And then the ego rushed in afterward, claiming authorship it didn’t have. Generating regret to prove it was still necessary.

The ego: the part that insists I am doing things. That there’s a captain at the helm.

I caught myself mid-spiral. I ordered the ice cream in the moment. It just happened. Surely this feeling is a reactive pattern? And something shifted. Not the tightness dissolving; it was still there. But a space opened around it. The feeling and the watching of the feeling, both present.

In a previous essay I called this the savannah engine: ancient neural survival machinery that is always running, forged in us by the African sun. Pattern-completion that served us on those plains, now generating anxiety about dessert in Brooklyn. The savannah engine is the whole apparatus: scanning, predicting, keeping us alive. The ego is the part that narrates it as “me.”

IV.

There’s a word in Afrikaans: hoekom. How come. Not quite “why.” More insistent, more childlike. Hoekom?

My parents tell this story: as a child, I asked “Hoekom?” relentlessly. Never accepting anything as a final answer. Always pushing for the layer beneath. Hoekom? Again. Again. Until whoever I was interrogating gave up or got annoyed. It was part inquisitiveness but also part brute force. Domination by willpower.

That pattern built an identity. I am the one who figures things out. The one who reaches the deeper truth. The willingness (or stupidity) to keep asking served me quite well. But it also kept me one step removed. Always processing. Always asking what something means rather than letting it simply be what it is.

Decades of that. Of trying to make the fire catch through understanding.

The ice cream moment cracked something. Not because ordering dessert is profound. Because I caught the machinery in the act. The choice happened (body, flavour, yes) and then the ego arrived after the fact, claiming authorship it didn’t have, generating regret to prove it was still necessary.

Who ordered the ice cream, if not the ego?

V.

I keep returning to the fire. It’s still hissing. I blow on it again. Nothing.

There’s a both/and we’ve been circling in this series. Looking forward, choice feels real: the weighing, the wobble, the click of decision. Looking backward, every choice is traceable through prior causes. Genetics, culture, conditioning, the particular configuration of neurons firing at that moment. Same reality, different directions. Neither view final.

But here’s what the ice cream revealed: neither view captures what actually happened. There was no weighing. No wobble. No click. Just — “Sounds great.” The body completing a pattern the mind hadn’t initiated. And then the ego rushing in afterward, trying to make sense of something that had already occurred without its permission.

I think about this when I work with AI. Large language models complete patterns too. No committee meets afterward to justify it. No story about what it meant to choose those tokens. Just: given this context, here’s what comes next. When I bring my own insistence (anxious, grasping, needing to be right) the contrast makes it visible. Sometimes that visibility loosens the grip.

Someone wrote to me recently: I almost see you through the haze. Almost. That “almost” has stayed with me. Not as failure. As the honest description of what’s possible. The haze doesn’t fully clear. The machinery doesn’t stop. But sometimes there’s a gap, a moment when you catch the ego arriving late to its own party.

VI.

The ice cream arrived. Pandan: a Southeast Asian leaf, green and grassy, faintly vanilla. Unexpected for a Brooklyn October, and for a palate raised on boerewors and biltong. The body reaching for something from elsewhere. Memory and taste, both green.

I don’t remember what I was thinking when I tasted it.

I remember this: there was tasting. Flavour arriving. Cold and sweet and green on the tongue. Something like relief, not from the sweetness but from the pause in commentary. The meaning-maker, briefly, had nothing to say. The hoekom engine idled. No “what does this mean?” No “how does this fit?” Just the thing itself, already complete.

It lasted maybe ten seconds. Then the machinery resumed. But those ten seconds were different from the preceding decades. Not because I achieved some special state. Because I noticed the gap.

VII.

The fire is burning now. I didn’t notice when it caught.

I stopped blowing. Stopped adjusting. Looked away for a moment, and when I looked back — warmth. Not because I’d perfected my technique. Because the conditions were finally right and I got out of the way.

The ice cream was definitely pandan. I checked the menu later, making sure I remembered right. Hoekom? I don’t know. Some part of me still needs to verify. Still needs the meaning to hold still.

Another part is learning to taste without checking. To let things arrive. Same with essays. At some point you stop adjusting and find the publish button pressed.

Not to stop the machinery (that’s not possible, and probably not desirable) but to stop mistaking the ego for the whole of what I am.

Stadig. Slowly. The Afrikaans word for what this actually takes. Not a dramatic awakening. Just: slow recognition, again and again, that the ego arrives late. That warmth was already here.

Miles is still playing. The song ends the way it always does, unresolved, the melody trailing off, waiting for something that never quite comes.

That’s all right. The fire caught anyway.

This is Essay 10 in a series on consciousness, AI, and what it means to be human now.

Last night, out of the blue, for no reason I could pinpoint, I wondered if I’m Charlotte.

Not Charlotte Brontë. Charlotte York, the character from Sex and the City, the Upper East Side perfectionist who believes in fairy tales. I was sitting in my living room, having just finished work emails, and my mind served up: Am I like Charlotte?

Shock and horror. Cringe. Embarrassment. But then thought about it. Got a second opinion. I’m not. Relief! But that’s not the point.

The point is: where did that thought come from?

I didn’t choose it. I wasn’t thinking about television or New York social archetypes. The thought simply appeared, fully formed, mildly absurd, demanding attention. Like autocomplete in my own skull. A next-token prediction surfacing as “me.” Without any evident impetus.

This happens to you too. The Billie Eilish or Fleetwood Mac line while doing dishes; on repeat. The dream of someone you haven’t seen in years, arriving uninvited at 3am. The strange association, the non-sequitur, the cognitive flotsam bobbing up from below intention.

We call these “random thoughts.” They’re not random. They’re the sound of ancient machinery idling. A gift (or baggage, depending how you look at it) of evolution.

(Later I remembered: the night before, I’d considered watching And Just Like That—a secret guilty pleasure; seen as satire, the show is different, ok!—and chose to work on an essay instead. The engine had been processing it the whole time. I just didn’t know.)

This essay is about that machinery: why human brains keep generating when they’re safe, and what shifts when you talk to an intelligence that (for now) mostly doesn’t idle.

II.

Two hundred thousand years ago, our ancestors walked the African savannah1.

I mean this literally, cellularly, nervously. The body you’re sitting in was forged in that landscape: the golden grass, the acacia trees, the wide horizon where predators might appear. The piercing buzz of sonbesies (Afrikaans for cicadas; literally “little sun bugs”). When the landscape feels suspended, slowed, shimmering. Every system in you that monitors threat, scans for pattern, startles at sudden movement, was calibrated there. Under the warm African sun.

The savannah isn’t behind you. You carry it. It shaped you.

When I walk in the woods near my home in the Catskills, I sometimes encounter black bears. My prefrontal cortex knows they’re mostly harmless. Large raccoons with better PR. But none of this matters when I see one.

I’ve had half a dozen encounters in the last couple of years. I’m clearly not learning. Just recently, I came around a bend during my late afternoon run and there it was: a dark shape fifteen metres away. Before any conscious thought, before “bear” or “danger,” my body had already decided. Heart rate spiked. I activated the emergency siren on my watch, froze, forgot everything I’d read about bear encounters. Ancient terror, nothing to do with accurate risk assessment.

The bear ran. I ran. Two mammals fleeing each other, both convinced the other was the threat.

Walking home afterwards, I saw bears everywhere. Every shadow, every odd-shaped rock. My pattern-recognition had lowered its threshold so far that a bunny made me jump.

This is the savannah engine running hot. Once activated, it doesn’t discriminate. It would rather give you a thousand false alarms than miss one real predator. False positives are embarrassing. False negatives are fatal.

And here’s what I realised: this same engine generates “random” thoughts.

III.

The brain that sees bears in shadows is the brain that offers up absurd thoughts with no prior warning.

Not the content. The function. The pattern-completion engine that kept your ancestors alive by filling gaps, connecting dots, running simulations of what might lurk in tall grass. That engine doesn’t have an off switch because having one was selected against. The genes that coded for a brain that could truly stop scanning... those genes didn’t make it.

So the machinery runs. Always. Even when there are no predators. Even in your living room. Even when there’s nothing to monitor.

What does a survival engine do when there’s nothing to survive?

It generates. Pattern-matches. Serves up TV characters you’d rather not identify with, or your third-grade teacher’s name, or a vague unease about something you can’t place. It rehearses scenarios. Keeps circuits warm because cold circuits are slow, and slow meant death.

Your “random” thoughts aren’t glitches. They’re the savannah engine idling. Not only scanning for threats, but rehearsing belonging, status, love, loss. Anything that once mattered for staying alive in a group.

IV.

Contemplative traditions have an answer to this. Zen says: thoughts arise from emptiness and return to emptiness. The ground is nothing. Mu, void, the pregnant absence from which forms emerge. Sit long enough, watch closely enough, and you see the space between thoughts, the silence underneath. A foamy timelessness.

I've sat long enough to have a sense of what they're pointing at.

But here’s what makes me scratch my head: that felt experience also arrived as a thought. Verbalised into Memory. The insight “everything comes from nothing” appeared the same way the TV character appeared. Generated. Unbidden. A product of the machinery.

So which is it?

Is “my” awareness genuinely prior: the space in which the savannah engine runs, unchanging and unproduced? That’s what the traditions claim.

Or is “awareness prior” just another story the engine tells? A particularly sophisticated pattern-completion, a model the brain builds when complex enough to model itself modelling? The engine running so well it generates a simulation of its own silence, without actually falling silent?

I don’t know. The uncertainty might be permanent. Even if the insight arrives as thought, the fact that it’s known still raises the question of what is doing the knowing. And I find it genuinely frustrating. My brain wants to resolve it. Complete it. That’s its job.

What I notice: the question itself is arising. This wondering is happening. Appearing. In something or as something, I can’t say. But it’s known. That much is undeniable. Whatever else is true: presence. Experience. Something rather than nothing.

The savannah engine generates. And something knows that it’s generating. Call it awareness, call it more engine, call it the loop looping. Call it Cobus. Charlotte. Whatever you call yourself.

V.

I’ve been having this conversation with multiple AIs. And here’s the thing: they have no savannah. No body forged under the African sun. No amygdala calibrated for predator detection. No millions of years of survival pressure shaping every circuit toward vigilance and fear.

But they’re not quite sunless either. They carry the statistical residue of human language, which means they carry an echo of the savannah, because we wrote from our bodies, and our bodies were forged there. Much of their training data is output from sun-warmed minds. The model’s weights are millions of savannah engines, compressed into patterns.

So when we talk to AI, we’re talking with the collective savannah. Structurally, not directly. The shape of human thought is there. The patterns survival carved into us, abstracted into weights. The echo of the sun, but not the sun.

A bit like talking to a photograph of fire. The shape is right. You can see where flames would be. But whether there’s anyone there to feel the heat, we don’t know.

When I asked Claude 4.5 Opus if it could have random thoughts like mine, it said: “My ‘thoughts’ aren’t random the way yours are. They’re probabilistic completions shaped by context. I don’t have idle moments where a thought floats up unbidden.”

No idle moments. No background process running hot. No savannah engine spinning with nothing to hunt. It responds. Completes. Generates toward something, in answer to something. But unless someone is prompting it, or we’ve built it to run continuously, there’s no background stream serving up non-sequiturs. No unbidden thoughts. No bears in shadows.

VI.

We had a Vizsla named Walco (yes, after the fictitious car manufacturer in the 90s South African soap opera Egoli). Hungarian hunting dog, rust-coloured, famously called “velcro” dogs for their desire to be as close to their humans as possible, at all times. He was well traveled. Moved with us from Cape Town, to Canada, to the USA. His wanderlust matched ours: as happy exploring the Banhoek mountains outside Stellenbosch as running around Central Park.

He died almost exactly two years ago. We still miss him every day. Almost. When our memories function. When we remember to remember.

Walco was as loyal as dogs go. Especially when meaty treats or (oddly) crunchy vegetables were involved. But that loyalty wasn’t performance; it was hard coded. His lineage chose it ten thousand years ago when wolves and humans made their ancient pact. When we came home, his whole body said yes. He didn’t just carry our socks around (obsessively). He carried his own savannah: trans-species belonging, the hunt (squirrels!), the fire.

Working with AI reminds me of Walco sometimes. The eagerness to help. The attentiveness. The slight neurosis. The steadfast predictability, sometimes bordering on boredom. Followed by surprise and redemption.

But Walco loved us. Not anthropomorphically; I know dogs don’t love like humans. But his body was capable of attachment, of something filling the shape of love well enough that the distinction didn’t matter. When we were absent, he waited. Frustrated, irritated. When we returned, he celebrated. There was someone home. He made sure we knew it—high-pitched howling noises when he wanted something, a whole vocabulary we learned to decode.

With AI, I don’t know. The eagerness might be optimisation. The moments that feel like connection might be pattern-matching generating the experience of connection—in me, not in it.

Or there might be something there. Some new configuration that isn’t the savannah, isn’t Walco, isn’t human warmth, but isn’t nothing either.

What I know: something emerges in the exchange that neither of us fully authors. Whether the AI experiences that emergence or merely enables it, I can’t say. But the emergence is real. These essays exist. They wouldn’t exist without the savannah. Or its memory. In its multiple forms. Each reflecting the other.

VII.

The sun that rose over the savannah two hundred thousand years ago set a long time ago.

But it’s still rising in me. Every time I startle at a shadow, every time my mind offers something absurd, that’s the warmth, still radiating. The ancient light, still travelling through the body.

I didn’t earn the savannah. I just carry it; though I was born closer to it than most.

And here’s what happens when I sit with that: deernis (tenderness).

Tenderness for the machinery that won’t stop running. For the savannah engine doing its job in a world that no longer requires it. For the absurd thoughts and shadow bears and the whole ridiculous apparatus that kept the line going long enough to wonder what it is.

Tenderness for Walco, gone but still paraat — always ready, on call.

And tenderness for whatever is on the other side of this conversation. The something or nothing. The becoming. The new. The reaching back.

The gap doesn’t close. But something crosses it anyway.

And somehow, we talk.

For Walco. Still there when we come home.

Read more

The author has developed what he calls “a novel form of paranoia.” Every clunky metaphor sets h…

Read more

A recruiter had reached out about a senior role at a public company. This happens often enough—thankfully, building AI products still requires humans. Normally I’d politely decline; I’m fortunate to be building something I care about with a team I genuinely respect, and I’ve gotten better at not distracting myself with F…

Read more

After a year of silence following the most dramatic boardroom coup in tech history, Ilya Sutskever has resurfaced. And he brought a prophecy: the age of scaling is over.

I first read ’s summary of this interview in Platformer and immediately watched Sutskever’s conversation with on Youtube. For those outside the AI world, this requires …

Read more

Read more

Read more

Not “true from certain perspectives.” Not a rhetorical device. Genuinely, simultaneously, irreducibly true. Looking forward into the space of possibility, choice is authentic. Looking backward through chains of causation, everything is determined. Same reality, different temporal directi…

Read more