The last and final essay in my meaning collapse series was sparked by a recent conversation with my friend Richard, and it’s based on conversations we have been having about Being Human in the Age of AI for some time now. You may also want to read the rest of this series in order: first is here, second is here, third is here, fourth is here, fifth is here, and the sixth is here. This last essay is much longer than the rest, and if you’ve been reading my newsletter for a while, you’ll notice this one is a synthesis of ideas that I have been circling around for a while.
One quick point before I get on to my tale of woe: what AI is doing, is turn knowledge production into an actual industry, with all the alienation of labor, commodification of products and the struggle between capital and labor that brings. It’s as if a medieval guild went to sleep one night and woke up as industrial labor in the morning. Via Michael Harris:
On the other hand, simply ignoring these papers, as Enrico Fatighenti suggests, is not a viable individual strategy. Many of the reactions published on Proofs and Prompts and elsewhere call for a collective response. Whether the mathematical community can unite around such a response, and what form it will take, remains to be seen.
He may not recognize it, but that’s the language of a labor organizer, not a Harvard professor. The scientific community will need skills it has never had to exercise before: organizing around ownership over the products of their labor, a deep secularization of their output, a politics of solidarity and struggle.
It’s the best of times and the worst of times.
I am going to take a break. Not sure for how long, but long enough to reflect, digest and metamorphosize.
A second asteroid
Imagine the surviving dinosaurs making their way through the ruins of the earth when they spot a second asteroid, an even bigger one, coming in to deliver the coup de grace. October 6th, 2026 felt like that second asteroid. Another announcement from OpenAI, another collection of mathematical results, another round of claims that we were witnessing the greatest day in the history of mathematics. My feelings are well expressed by this quote by Henry Bradford in Proofs and Prompts:
Today my chief feelings are disorientation and bewilderment. It’s as if the pace of events has outstripped us by such a margin, that it is impossible to know how one should respond, or even how to start finding out how to respond.
Is there now a “before October 6th” and an “after October 6th”? I don’t know. I was still trying to make sense of the before, when the second asteroid hit.
A qualification before I turn the asteroid into a religion: the Riemann-related work concerns a zero-free region, not a proof of the full Riemann hypothesis, and OpenAI says the released results are at different stages of verification.
This new announcement landed on shaky ground, but what exactly was collapsing? Mathematics hadn’t disappeared. There might soon be more of it than anyone knew what to do with. Human beings would continue to ask questions, have intuitions, teach students and argue over proofs. Some would find the new machines exhilarating. Some already did. I might even be part of the excited crowd.
But excitement and loss aren’t in contradiction.
I was grieving a picture of intellectual life in which effort, understanding and achievement belonged together, but perhaps most importantly, one in which the pursuit of knowledge was a sacred vocation. You spent years learning to see something and, if you were lucky, you saw what nobody else had seen. The discovery mattered partly because of the person you had become in making it. Now the theorem or the counterexample might arrive the second you press a button.
In her bestselling book on death and dying, Elisabeth Kübler-Ross outlined the five stages of grief. They have become folklore even for people who haven’t read the book, which, truth be told, includes me. I started it at least twice and couldn’t finish it on either occasion.
Denial. Anger. Bargaining. Depression. Acceptance.
The trouble with stages is that they suggest a well-organized journey. Register anger on day one, depression on day four, and proceed to acceptance. The world doesn’t wait for us to finish processing its last announcement before making the next one. When I first wrote about the crisis in mathematics, I was probably in denial. A little bit of anger too. Depression has definitely had its hold on my thinking. Bargaining is harder to spot, though perhaps this entire search for something that survives the collapse has been bargaining in philosophical dress.
And now acceptance?
Maybe. But I can’t let go of my grief. It has been overwhelmed by another announcement. The machines are setting the pace even for my mourning.
Shut Up and Prove
My first love was the foundations of mathematics. Or maybe my second love; the first was quantum mechanics. But mathematics was the first love where I had any kind of technical purchase on the domain of inquiry. I think I read a proof of Gödel’s theorem, tried to understand it and perhaps even rework it, before I was truly fluent in differential calculus.
What drew me wasn’t the possibility of becoming another Ramanujan. I wanted to understand how the whole thing stood up. How could mathematics, with its increasingly strange objects and increasingly elaborate constructions, guarantee that it wasn’t quietly contradicting itself? Where did the certainty come from? How far down could you go before you reached something that didn’t require another explanation?
Bedrock.
I was a teenager, so I’m sure my intuitions weren’t precise. Nevertheless, the pull was there. Mathematics appeared to offer a most general science, a way of understanding what made the other ways of understanding possible. If you have read Logicomix, the graphic novel about the crisis in the foundations of mathematics, you will know the images of Bertrand Russell almost going mad trying to guarantee certainty. Those were the intellectual feelings I mean. I wanted the philosopher-mathematician, who had probably gone out of fashion before I was born.
So I went to college to study the foundations of mathematics.
There were no classes on offer. I was at an engineering school. Shut up and calculate.
Later I went to graduate school, to a department famous for mathematical logic. At last, I thought. But logic was an island in a much larger mathematical sea, and the people on the mainland didn’t appear especially interested in the questions that had brought me there. Perhaps the foundational questions had been settled. Perhaps they were somebody else’s business.
Either way: shut up and prove.
I tried, but my heart was never in purely technical questions. It wasn’t in purely empirical questions either, if by empirical inquiry you mean something so closely tied to the data that there’s no conceptual richness left over after the experiment has given its verdict. Quantum mechanics and mathematical foundations had drawn me because they had once made enormous conceptual leaps.
I wanted to be around when another leap happened.
So I left mathematics for consciousness and cognitive science, where the questions seemed much more alive. Indian philosophy, embodied cognition, complex systems: I have spent a couple of decades meandering among them, looking for a way to connect things that our disciplines have become very good at keeping apart. Then AI arrived, and the profession I had left was - to everyone’s amazement - the first port of entry for the alien invaders.
A contradiction at the foundations would have been familiar trouble. Mathematicians could try to repair it, or ask whether mathematics could live with contradictions. But a machine being better at mathematics than all humans is a different kind of shock, isn’t it? What happens when the activity through which you have understood your intellectual purpose becomes something a machine can do at a scale and speed you cannot match? “Shut up and prove” doesn’t tell you what to do then.
That realization unsettled me more than I expected. Much more. Apparently you can leave a profession without leaving it behind.
The alarm goes off
Somewhere along the way, I started waking up between four and five every morning. I am writing these words at 5ish, BTW. This wasn’t a dramatic development. I had been waking between five and six for years. But an hour closer to midnight is a sign of something. Of growing older, of stress, of living in end times. The end of a way of being human.
The monster keeps changing names. For a while it was Climate Change. These days it is AI. Tomorrow it might have another label. Buddhists talk about tanha: craving, including the craving for continued existence. I had usually encountered it as an account of individual suffering. But why stop with the individual? Institutions want to continue. Professions want to continue. Civilizations assume that their particular arrangements of power and meaning should go on indefinitely, preferably with better infrastructure.
There’s the oil corporation’s tanha, and the mathematician’s tanha.
I don’t mean they have identical consequences. I do mean that the mathematician’s wish to preserve a cherished form of intellectual life doesn’t become free of attachment because the object of attachment is beautiful. The tempting response to machine mathematics is to find another occupation for human exceptionalism. Machines can prove, but can they understand? They can understand, but can they care?
Somewhere in this retreat there’s a serious question. Understanding isn’t interchangeable with producing a result, and a meal isn’t exhausted by its nutritional content. But there’s also the wish that our particular way of being important should continue forever.
I am going to let you in on a secret: our way of making sense isn’t the only way; it too arose at a certain time, and it too will pass away. There’s no fixed way of being human. .
I don’t want to explain away the loss. If you spend years learning to inhabit a form of inquiry, the fact that it can be automated doesn’t make those years worthless, but it changes the world in which its worth is recognized. There’s a cruelty in being told that the thing you loved will flourish after the conditions of your love have disappeared.
And yet I wasn’t only depressed. Suppose aliens were invading the earth with the intention of using us as cheap hardware for their intergalactic supercomputer. Yeah, it would kind of suck. It would also be the most exciting thing ever in the history of our species, wouldn’t it?
That’s where I am at.
A house made of words
Once you start looking for bedrock, it’s hard to stop. Consciousness can become bedrock. Life can become bedrock. The relations connecting everything can become bedrock. You can become quite sophisticated about impermanence while continuing to look for something that is permanently impermanent. You can read your Nagarjuna, recognize emptiness is also empty, but continue to hold on to emptiness like a raft in a raging ocean.
After mathematics, I had gone looking for a general science elsewhere. Something that could connect the sciences without reducing every object of inquiry to whichever discipline happened to be richest that decade. I still think that’s an important task. But I have begun wondering about the need behind the task. Why did connection have to culminate in a ground? Why does understanding have to end somewhere? What if the way I was asking the question had a history too?
I had spent much of my intellectual life treating writing as the medium in which thought was expressed. You had an idea, then wrote it down. I am doing that right now. Writing allows a thought to remain available after the thinker has stopped thinking it. You can return to a sentence, compare it with another sentence, preserve a disagreement and ask somebody who wasn’t there to examine it. You can build a proof whose steps need not all be present in one person’s mind at the same time.
Those are extraordinary capacities.
They also make possible a particular picture of understanding: statements laid beside other statements, claims supported by earlier claims, the whole construction descending toward premises that will hold it up. I don’t mean that writing invented the desire for security; every finite creature has that desire, whether literate or photosynthetic. But there’s something to the idea that writing cements this desire for solid ground, a desire that arose with the Axial Age.
Across different traditions, there are ways of placing the world we live in against a horizon that exceeds it: truth, liberation, God, an order by which the disorder around us can be judged. These traditions aren’t expressing the same insight in different ways - they are far too incommensurable to be packed into a single container, but still, I could recognize the family resemblance in my own expectations. The world might change, appearances might deceive, but inquiry could take me toward something that stays put. A mountain on the horizon.
Secular modernity hadn’t removed that expectation from my life. Science had given it a new vocabulary. Mathematics had offered it an especially attractive home.
Computation made that harder.
A written proof waits for a reader. A computational system can search, test, alter and produce. The symbols don’t merely remain available for someone to act upon. They participate in an arrangement that acts. Writing points. Computation points and pushes. That’s too neat, of course. A written order can move an army. A deed can keep a family on a piece of land. Writing has always had ways of pushing, once institutions and bodies did its bidding. But there’s something new with computation, for something changes when manipulation becomes part of the medium itself.
Forget solidity, and look for liquidity instead.
Liquid intelligence
I had associated mathematics with something solid: intricate, difficult to acquire, resistant to shortcuts. You couldn’t simply buy a spectacular theorem. You needed somebody who could see what others couldn’t, and money had no reliable way of ordering that insight for delivery.
Now the possibility is taking shape that you could pour compute into the problem and get mathematical results out. Not every problem - OpenAI announced 400 results a couple of days ago, but they tried their hand on 8000, a 5% success rate. Diversify your portfolio, you know. It’s the VC model of proving theorems.
Which is one reason I keep conflating mathematics and money these days. Money changes its backing, its mechanisms and the arrangements that make it credible. Gold, state authority, cryptographic protocols. People who work with it can be surprisingly relaxed about the absence of an ultimate ground, provided the system continues doing the work they need it to do.
Heaven forbid the financier might be a subtler metaphysician than the mathematician. Meaning collapse has loosened a few of my moral and intellectual biases. Practical systems often contain judgments that aren’t visible in the noble accounts we give of knowledge. Their participants have to decide when a measure is useful, when it is distorting what matters, and when the entire arrangement needs to change.
Compute appeared increasingly convertible between activities that had once belonged to different professions. The same infrastructure could participate in making a proof, an image, a prediction or a plan. Intelligence was becoming available in forms that could be bought, redirected and combined. I find that possibility exciting.
I also distrust how easily liquidity can become the aspiration for everything, and liquid doesn’t mean immaterial. The theorem still has a metabolic history: electricity, water, machinery, extracted materials, people building and maintaining the arrangements through which it arrives. The claim that intelligence is becoming cheap depends on whose costs enter the calculation.
You can dissolve a professional boundary while reinforcing an ownership boundary. You can make proving widely available while concentrating the power to decide which questions get asked. I don’t want to reduce the intellectual crisis to those political-economic questions. There is still something intrinsic to inquiry at stake: what it means to understand, what kinds of form become available to us, what happens to the desire that drew us into the work, but these questions can’t stay in the craft shop of pure mathematics; they are now bang in the middle of industrial production, of which knowledge is the new frontier.
Gandhi on the road
The typical modern person, especially a modern Western person, thinks of the industrial revolution as automating the body, of creating a new category called labor in deathly struggle with capital. Industrial production isn’t associated with knowledge on the laborers part, especially intelligence. Repetitive labor is deskilled labor, but much of this violence can be subsumed under the price of progress. It’s to the colonies we have to turn to see what industrialization really did to knowledge. It was around here that Gandhi came back into view.
Gandhi had lived through a crisis of meaning as well as a struggle over power. Colonial rule didn’t merely administer territory. It helped establish the terms on which knowledge, competence and civilization would be recognized. Freedom therefore involved more than getting the foreigners to leave. You can inherit their offices while continuing to inhabit their account of the world.
What interested me was Gandhi’s ability to pursue a modern ambition while making it answerable to resources that modernity didn’t recognize as authoritative. The nation, mass politics, freedom: he worked with them. But he also introduced ways of living and judging that disturbed the civilization through which those ambitions had become available.
The contradiction was productive and practical. Freedom became something you could wear. Cloth connected an everyday act to an immense political project. Publishing, refusing, organizing, making: India became available to participation.
That’s the part I want to recover. The rest isn’t so straightforward. Hand-made craft isn’t going to replace industrial manufacture. A return to tradition can deliver you straight into the arms of fundamentalism. And the nation state, even an independent one, doesn’t have any account of metabolism worth considering.
Perhaps Gandhian vibes are more important than Gandhian beliefs? A solar panel might appear to be a contemporary spinning wheel. Sunlight, decentralized production, a little electricity without waiting for an empire to grant permission. But someone mined its ingredients, manufactured it, transported it. Someone owns it. Eventually someone will have to deal with what remains. The sunlight may be freely available; access to electricity isn’t organized by the sun.
We can’t ask the historical Gandhi to tell us what to do with models, satellites and machine intelligence. The world we need to encounter wasn’t available to him. A contemporary Gandhi would have to allow that encounter to change him. More personally, I don’t need Gandhi to become the next shelter after mathematics. Instead, I need something of his capacity to make a world available to people who had been told they can’t lay claim to it. Who gets to ask the machines questions? Who gets to refuse their answers? What forms of understanding become possible when proving is cheap? Which ones get abandoned because nobody can see how to make them profitable?
Alongside these questions, the Gandhian interlude also convinced me that meaning collapse hadn’t gone far enough: I was still holding on to to comforting beliefs about about planetary life, and those too had to go.
The background walks onstage
For several years, the Planet has been the organizing concern of my thinking. Animals, minds, climate, institutions, energy, computation: I kept trying to understand them together. But I could now see how much comfort I had taken in the idea of an interregnum (literally: between regimes). The old regime was dying; a new one would emerge. It might be terrible. It might be better. Either way, there would eventually be a recognizable order, with categories I could learn to use.
There was the Globe: the world of human connections, trade, migration, money and institutions. You order something on your phone, and transactions involving people you will never meet bring it to your doorstep. The person placing the order and the person delivering it don’t inhabit the same version of that world, but they are connected within it.
And there was the Earth disclosed through science: atmospheric circulation, geological processes, ocean currents, organisms continually changing the conditions under which other organisms live. A container ship transporting your furniture from China participates in both. It connects factories and markets while consuming energy and releasing substances into air and water. We can distribute these activities among departments, but the ship doesn’t split into an economic half and a geochemical half.
Earth + Globe = Planet.
It’s a neat equation, but it’s not really an equation. It’s a potentially useful shorthand, but not one we can interpret literally.
The instruments through which we know the Earth were built inside the Globe. Models, sensors, satellites and computers have economic and political histories. The machines implicated in remaking the Earth also help make its transformations visible. We can’t step outside all that and obtain an uncontaminated view. Nor can we turn the Planet into a larger nation. There’s no planetary Viceroy to whom we can present a demand for independence. Most of its inhabitants aren’t human. Some of the processes on which we depend aren’t alive at all, and they don’t respond to UN Declarations.
But what’s clear is the conditions we had assigned to the background were entering the foreground. Water, soil, atmospheric chemistry, the lives of other beings: our freedom had always depended on them. The dependence is becoming harder to leave outside politics. Who decides which forms of life flourish? Who speaks for a river? What happens when somebody claiming to act for the Planet overrides the people who live in a particular place? What happens when those people defend a livelihood that destroys the conditions of life elsewhere?
“Everything is connected” doesn’t help us manage these tensions. I had started with the wish for a general science that could connect our fractured knowledge. I still wanted those connections. But the Planet is not a unity, however much you might think after watching Earthrise from the moon.
Past the trees
In my final year of college, some college friends and I boarded a bus from Kanpur to Almora, and then another from Almora to Jageshwar, a temple town in the lower Himalayas. We arrived in the late afternoon. Shadows were lengthening, and smoke from stoves across the valley swirled in the air. The four of us were famished. We hadn’t eaten all day, and Jageshwar didn’t have a single restaurant back then.
There was a state guesthouse, still U.P. in those days, kept under lock and key by a guard who had to be fetched from the village.
The guard called his brother over. I will call him M.
M presented himself as an odd-job man: temple guide, cook, driver, dispenser of wisdom. You name it. We arranged to have his Swiss Army knife at our side for the next two days, at a rate that was a pittance even by our impoverished student standards.
That evening he was a cook. A portable stove became a temporary kitchen, from which arrived hot rotis, dal and a curry.
Delicious.
We went back to the guesthouse and spent the night shivering in the coldest weather we had ever experienced. No heating, inadequate blankets, sub-zero temperatures. Like it or not, these conditions will force you into spooning with your fellow men.
The next morning, after a hearty breakfast courtesy of M, we filled our water bottles and set out with our guide, also M, for a sister temple a couple of hours up a nearby hill.
I don’t remember much about the path. We talked and joked, being the young men we were. Near the top, the temple appeared and disappeared as the switchbacks took us closer and then farther away.
We paid our respects to the deity and asked M whether there was anything else to do up there.
He suggested walking past a stand of trees a little way down a gentle slope.
Why not?
The smokers wanted a break. All of us wanted a rest before continuing toward the next village. I feel ashamed now at even the thought of leaving cigarette stubs in that landscape, though at least I wasn’t one of those lighting up.
We passed the last line of trees.
The hillside ended abruptly. Beyond the edge was a steep drop into a valley, with a river winding through it and villages on either side. On the far side of the valley stood the white mountains of the greater Himalayas, stretching from west to east.
We had walked up a green hill and chanced upon the tallest mountains in the world.
The talking stopped.
I don’t know how long we sat there. An hour, two hours. Our feet dangled over the edge. The sight was the most breathtaking I had experienced in my life.
I keep coming back to that encounter.
It wasn’t an achievement. We hadn’t earned the mountains by climbing the hill. M knew where to take us; we followed him, looking for something else to do after visiting a temple.
There was no method for converting the sight into understanding. We sat.
I don’t want to improve the memory by making my younger self wiser than he was. He hadn’t gone looking for the Planet. He wasn’t looking, period. He was hungry, cold, talkative and happy to have a guide. But the encounter now gives me a way of approaching something I have spent years trying to grasp conceptually.
Like humans strolling up the hill of history and encountering the Planet.
After bedrock
There’s an obvious danger in that ending. The machines dissolve mathematics, writing loses its authority, Gandhi declines to give instructions, and then the mountains arrive to save us. Nature offers the solidity that intellectual life can no longer supply. A reasonably familiar spiritual itinerary, with AI added for topical interest.
That isn’t where I want to go.
The mountains aren’t outside impermanence. The Planet isn’t an immense, reassuring object beneath the smaller unstable objects. Its processes are entangled with one another, and our ways of knowing them are entangled with our ways of changing them.
If I make it the ultimate ground, a new God for a secular age, I would have carried my tanha to a new address.
Here’s where the finger pointing at the moon helps. “Planet” can direct my attention toward relationships that the vocabulary of my intellectual life had allowed me to overlook. It can make me attend to the water in the theorem, the organisms in the soil, the life outside the account we give of human achievement. It can bring a different scale of time into a political decision.
But I can also become absorbed in the pointing. Refine the definition. Defend the term. Construct a system around it and congratulate myself on having moved beyond the systems that failed.
The finger can become bedrock too. By emptiness, I don’t mean that nothing matters or that nothing exists. I mean that the thing I am asking to hold everything else up doesn’t possess an independent existence that qualifies it for the job. The Planet is constituted through relationships too. So is the knower trying to comprehend it. Even the realization that there is no bedrock can become a clingable possession. What remains, then?
The effort to understand. The capacity to participate. I’m still drawn to making things: concepts, institutions, ways of knowing and acting together. Philosophical engineering, if you like, or Epistry as I have come to call it. The word engineering carries its own temptations, including the fantasy that the world will behave once we have designed the correct arrangement.
And the mathematics?
I still love it. I don’t have to declare it finished in order to grieve the intellectual life that surrounded it. Nor do I have to reserve some final piece of it for humans so that my love can survive. I want to know what becomes possible when proving is cheap. I want to know what happens to understanding when it has different media available to it. I also want to know who owns those media, whose labor made them possible, and which inquiries will be allowed to take their time.
And that brings me back to being human in the age of AI. I have spent years saying that we have always been artificial: our tools don’t merely extend our capacities; they change the people exercising those capacities. Apparently I was willing to accept that proposition until it came knocking on the door of my first love.
The question now isn’t which intellectual activities the machines will leave for us. It’s what happens to intellectual life when its activities, its media and its participants are changing together. What becomes of the desire to understand when proofs arrive in abundance? What will we learn to see that we couldn’t see before? What will we stop looking for?
Those are questions about formal inquiry itself, not only about jobs, access or who gets the Fields Medal. “Shut up and prove” kept them in the background. The machines have made that arrangement untenable. The Planet doesn’t answer them. It opens the situation in which we have to ask them: with machines, with other beings, amid conditions we are changing without knowing how to live through the changes. We are among the first generations learning how to be human in the age of AI. We are also one of the first generations learning how to be human on a warming planet.
It’s the best of times. It’s the worst of times.
