The treadmill is, when you stop to consider it, a fairly ridiculous machine. We take electricity, an electric motor, several hundred pounds of steel and plastic, a control system, a glowing screen, and a considerable portion of expensive interior real estate, and arrange all of it so that a human being can walk or run for an hour without arriving anywhere. For most of human history this would have seemed somewhere between extravagant and insane, because our ancestors spent an enormous portion of their ingenuity trying to reduce the physical exertion required simply to remain alive. Water needed carrying, fields needed plowing, wood needed chopping, clothes needed washing, goods needed hauling, and nearly every improvement in material life meant finding some way to spare human bodies a little of the punishment. A person who had spent twelve hours bringing in a harvest did not end the day with a vague sense of guilt that he had failed to close his rings. We built machines to relieve ourselves of all that, and then, having succeeded rather spectacularly, built another class of machines to put some of the exertion back.
Virginia Postrel and Charles C. Mann have been exploring precisely these hidden transformations in the Abundance Institute’s Everyday Abundance project. One episode traces the electric washing machine and the collapse of a regime of domestic labor that had consumed astonishing quantities of human life; another examines the invention of exercise as a normal social practice, something that would have baffled most people in the nineteenth century because strenuous physical exertion was already unavoidable. Their larger series proceeds through similarly ordinary things—recorded music, toothpaste, the innumerable technologies that have become so embedded in the background of life that we have forgotten they are technologies at all. Abundance changes what activities mean. Washing remains possible after the washing machine, but washday disappears as an organizing fact of domestic existence; physical exertion remains after mechanization, but it can become training, recreation, sport, even a peculiar form of leisure. Necessity recedes and an old human activity reappears under a different description because the person now stands in a different relation to it.
I think this is a much more interesting place from which to begin thinking about artificial intelligence and work than the now-familiar question of which jobs the machines will take. Of course that question matters too, particularly to anyone whose livelihood is attached to one of them, but it assumes that the great effect of a new technology is to redistribute an existing inventory of tasks between humans and machines. History is considerably stranger. Technologies of genuine consequence alter the background conditions under which those tasks, institutions, and identities made sense in the first place. The automobile did not create a more efficient horse economy, recorded sound did not simply give musicians a larger concert hall, and electrification did more than make existing household labor somewhat faster. New schemes of living became possible; old ones faded so thoroughly that later generations had trouble imagining why anyone had organized life that way at all.
This, at its strongest, is what I take the Abundance Institute’s wager to be about. Its formal policy work is concerned with creating the cultural, regulatory, energy, and infrastructural conditions in which emerging technologies can actually develop and diffuse; its AI work treats intelligence as a general-purpose capability whose importance lies in what happens when expertise that was historically expensive becomes dramatically cheaper and more available; its cultural work looks backward to recover how yesterday’s strange innovations became today’s invisible normality. These strands belong together. Abundance is not simply a bigger pile of outputs. It is what happens when a formerly binding constraint loosens enough that people begin doing things they previously could not do, institutions become possible that previously could not exist, and eventually a new generation mistakes the resulting freedom for the natural order of things.
But what sort of creature receives that freedom?
The thing that keeps asking
This brings me, as things often do around here, to Bernard Lonergan. Near the beginning of Insight, before the enormous architecture of the book has really begun to rise, Lonergan pauses over a fact of experience that is so ordinary we rarely think to make anything of it:
Deep within us all, emergent when the noise of other appetites is stilled, there is a drive to know, to understand, to see why, to discover the reason, to find the cause, to explain. Just what is wanted has many names. In what precisely it consists is a matter of dispute. But the fact of inquiry is beyond all doubt. It can absorb a man. It can keep him for hours, day after day, year after year, in the narrow prison of his study or his laboratory. It can send him on dangerous voyages of exploration. It can withdraw him from other interests, other pursuits, other pleasures, other achievements. It can fill his waking thoughts, hide from him the world of ordinary affairs, invade the very fabric of his dreams. It can demand endless sacrifices that are made without regret though there is only the hope, never a certain promise, of success.
The passage is beautiful, but it is easy to domesticate it into a pleasant observation about human curiosity. Lonergan means considerably more. The desire to know is the dynamic operator of his whole account of human knowing: experience gives something to be understood, the question arises, imagination and intelligence work upon the data until insight grasps an intelligible relation, and then intelligence itself presses onward into a different kind of question—Is it so?—which requires reflection and judgment. Human consciousness does not stop even there, because what is understood and judged gives rise to the further question of value—What am I to do? Is this worthwhile?—and judgment enters decision and action. What begins in wonder reaches, if we follow it all the way through, the responsible subject acting in a world. Lonergan’s familiar precepts—be attentive, be intelligent, be reasonable, be responsible—are descriptions of this movement from what is given through what might be understood toward what we affirm and what we make of ourselves in response. The desire is “unrestricted” because no finite answer closes the horizon; every achieved understanding can disclose another intelligibility still to be grasped, every answer can release what Lonergan calls the further pertinent question.
This is a much stranger anthropology than the one quietly presupposed by many discussions of technological unemployment. If human activity were principally the expenditure required to satisfy a bounded set of appetites, then the disappearance of scarcity should produce a corresponding disappearance of serious activity. Feed the creature, house it, entertain it, relieve it of burdens, and eventually it ought to come to rest. But Lonergan notices that when “the noise of other appetites is stilled,” another drive becomes audible. We ask. We wonder. An inconsistency bothers us. Something does not fit. A pattern glimmers without yet being clear. We read another book, run another experiment, take something apart, stare at a whiteboard, explain the problem badly to a spouse, wake at three in the morning with a connection we cannot quite hold, and occasionally experience the peculiar release of insight when the formerly unrelated suddenly hangs together. The answer satisfies us and, very often, creates a better question.
That phenomenon is difficult to fit into a theory in which work exists because people would otherwise prefer to do nothing. People obviously do prefer rest sometimes, as any honest anthropology must admit, but we also encounter mathematicians who give decades to problems with little promise of worldly reward, naturalists who catalogue species nobody asked them to catalogue, programmers who maintain open-source projects after their paying work is finished, woodworkers who spend weeks making objects available more cheaply at IKEA, runners who pay money to suffer voluntarily for twenty-six miles, and graduate students who have somehow concluded that translating an obscure fourteenth-century manuscript is an excellent use of several years. Economic incentives explain a great deal about the organization of human activity. They do not explain the whole animal.
If Lonergan is right, the question on the other side of abundance is therefore not merely whether human beings will still have something economically useful to do. It is what happens to the unrestricted desire to know when the cost of following a question falls.
And this is a considerably more exciting possibility.
Cheap answers and expensive questions
The most interesting claim in the Abundance Institute’s recent AI work is not, to my mind, its confidence that technological unemployment fears will once again prove overstated, although that argument has a long historical pedigree. It is the more concrete observation beneath it: practical expertise has traditionally been expensive. To obtain a bit of specialized intelligence about law, programming, design, medicine, accounting, engineering, research, or dozens of other fields generally meant finding a person who had spent years developing the requisite expertise and then acquiring some portion of that person’s scarce time. One reason organizations became large was that many kinds of coordinated expertise were difficult and costly to assemble. One reason individuals abandoned questions was that pursuing them crossed a threshold of money, access, knowledge, or institutional permission they could not clear. AI, to the extent that it makes useful cognitive capability cheap and widely available, changes the economics of the attempt before it changes anything else.
This is where Lonergan helps us avoid saying something silly about intelligence becoming abundant. Answers may become abundant. Formulations may become abundant. Analyses, code, translations, explanations, designs, simulations, candidate hypotheses, and whole libraries of the products of previous human insight may become available at negligible marginal cost. That is genuinely consequential. But inquiry is not the possession of an answer; it is an operation of a subject. Whatever view one ultimately takes of machine cognition, the human being encountering a situation still has to notice that there is something worth asking about, understand what question is really at stake, distinguish an answer that sounds plausible from one that is sufficient, decide whether further questions remain, determine what is worth doing with the result, and become responsible for the action that follows.
Indeed, cheap answers may make good questions more important. The student who can generate a competent essay in thirty seconds has not thereby acquired the judgment to know whether its argument is true. The founder who can produce twenty prototypes in a week has acquired possibilities faster than she has acquired the wisdom to choose among them. The scientist who can generate a thousand hypotheses has not escaped the problem of identifying the one that matters. When the production of plausible artifacts becomes easier, attention, discrimination, judgment and direction move toward the bottleneck. We do not cease needing intelligence because intelligence-like products are abundant; rather, the location of the difficulty moves upstream into the operations by which a subject determines what is worth understanding and downstream into the operations by which understanding becomes responsible action.
This gives us a better account of why the essay, the theorem, the program, the company, the garden, the scientific experiment or the handmade table can remain worth doing even after technology becomes capable of producing an adequate product by another route. The activity has an external term, certainly, but it also has an internal one. In understanding, judging and acting, the subject changes. Lonergan eventually makes this explicit in his account of the existential subject: by our free and responsible acts we make ourselves what we are to be. Work, at least in its richer forms, has always contained this reflexive dimension. We make the table and become a little more the sort of person who can see grain, feel resistance, judge an edge, wait for glue, correct a mistake. We write the argument and discover that the discipline of formulating it exposes what we did not understand. We care for a child and, by innumerable choices that produce no economically measurable artifact, become someone capable of a particular kind of patience, fidelity and love. The product matters; the act also has a product in the Subject.
AI makes that distinction harder to ignore because it can increasingly separate the external product from the human act by which we were accustomed to obtaining it. If the purpose of my daughter’s essay is simply to produce two thousand words on Hamlet, the assignment is in trouble. If its purpose includes forming attention, intelligence and judgment through an encounter with a difficult text, then the arrival of a machine capable of producing the words forces us to recover the reason the human act mattered. The same question will appear across professional life. Which of our activities were expensive ways of obtaining an output that can now be produced more cheaply, and which were practices in which the struggle to produce the output carried goods we will still want when the struggle is no longer economically necessary?
The washing machine removed washday. It did not establish that the human body had nothing left to do.
What technology actually changes
Lonergan has another idea useful here, buried deeper in Insight: schemes of recurrence. A scheme exists when a set of events forms a recurring pattern in which one occurrence supplies the conditions for another and the sequence, once established, tends to renew itself. Ecosystems work this way, economies work this way, institutions work this way, and much of ordinary life consists of innumerable overlapping schemes that are so stable we experience them simply as “how things work.” New possibilities enter history when conditions come together for a new scheme to emerge; once established, that scheme creates conditions for still other possibilities.
This is a better lens for technological change than the familiar inventory of jobs gained and lost. The washing machine mattered because it entered and reorganized a scheme of domestic life: plumbing, electrification, appliances, changing household roles, new uses of time. Recorded sound reorganized the scheme by which music was produced, distributed, encountered, owned and eventually created; as Everyday Abundance notes, the shift from music as a fleeting local performance to something recordable, reproducible and nearly infinitely distributable produced industries and artistic practices that an earlier world could scarcely have described. Cheap photography did something similar to the image. The interesting historical unit is not the individual task but the pattern of recurrence in which the task has a place.
That is why I want us to leave more imaginative room around AI than either the optimists or pessimists articualte. We keep asking which tasks in the current scheme can be automated and then drawing a line forward from there. But what if the important consequence is that the scheme changes?
Suppose practical expertise becomes dramatically cheaper while software creation, translation, research, design, coordination and as we’re already seeing in the startup ecosystem, some forms of physical fabrication become easier manufacture. Perhaps the important change is not that the lawyer writes a contract twenty percent faster, but that thousands of undertakings that previously required a lawyer, developer, designer, analyst and project manager to get through the front door can now be attempted by two people. Perhaps a capable individual begins to resemble a small organization, while a small organization acquires powers once associated with a large institution. Perhaps the minimum economically sensible size of certain firms collapses. Perhaps research networks emerge around people who would once have remained amateurs because instrumentation, computation and access to literature were too costly. Perhaps education becomes less concentrated in the first twenty-two years of life because a person can return repeatedly to serious learning as new interests and possibilities arise. Perhaps retirement becomes increasingly strange as a category when healthy older adults possess better tools than entire firms did during their first careers.
The wild possibility is not that everybody becomes a software entrepreneur. That would be a depressingly narrow use of abundance. It is that more human beings acquire the practical freedom to follow questions, form associations, try projects, learn disciplines, make things, care for people and begin enterprises without first passing through as many gates of capital and institutional permission.
Perhaps the most important abundance is an abundance of attempts.
That possibility fits the human subject Lonergan describes rather well. The unrestricted desire to know continuously generates questions, while practical intelligence asks how an understood possibility might be realized in concrete circumstances. Lower the cost of turning insight into experiment and a larger portion of human inquiry can escape the realm of the merely imagined. A question that once died because the equipment was too expensive, the expertise unavailable, the administrative overhead absurd or the prospective market too small may live long enough to become a project. Many of those projects will fail. Of course they will. Emergent probability does not promise that every possibility becomes actual; it describes a world rich enough in potential combinations that genuinely new patterns can emerge when their conditions are assembled.
An abundant society, on this account, would be less interesting because everyone owns more things than because far more people can bring possibilities under the discipline of reality.
Surplus is not yet abundance
There is, however, an uncomfortable fact hiding inside this argument. The appearance of extraordinary productive capacity does not guarantee that ordinary people become more free to act.
Lonergan spent an astonishing portion of his intellectual life thinking about precisely this problem in economics. His circulation analysis distinguishes a basic circuit, which produces goods and services entering the ordinary standard of living, from a surplus circuit, which produces the means of further production. The categories are relational rather than industrial: a machine is surplus because it enables production; the goods eventually yielded by the enlarged productive system enter the basic circuit as ordinary provision. During a period of expansion, resources flow toward building new capacity. If the expansion succeeds, however, the new capacity eventually has to deliver its fruit. The surplus phase reaches its intelligible term in a corresponding increase of basic goods and purchasing power. Lonergan calls the movement between the circuits the crossover.
The Abundance Institute spends much of its policy effort on what we might call the upstream conditions of a contemporary surplus expansion: enough energy, the ability to construct and connect infrastructure, competitive markets through which new capabilities diffuse, rules that allow experimentation, access to tools rather than their confinement to a few incumbents. Its AI argument similarly emphasizes the prospect that capabilities once available only to wealthy people and large institutions can reach ordinary households, workers and small firms. These are questions about whether productive possibility is actually allowed to become widespread capability.
Lonergan gives us the other half of the question. A society can possess tremendous surplus while failing to achieve abundance. Compute can multiply, data centers can rise, model capabilities can improve, capital values can soar and the productive apparatus can become almost fantastically powerful, while housing, health care, education, family time and the ordinary security required to attempt anything interesting remain stubbornly scarce. That is an economy rich in productive capacity whose crossover is malfunctioning.
The human test of abundance therefore cannot be how impressive the surplus circuit becomes. It is whether the new capacity crosses into basic life. Does expertise become cheaper? Do small organizations become more capable? Does a family need fewer hours of labor to purchase a decent standard of living? Does technological leverage accrue only to the owner of the machine, or also to the person who uses the goods the machine makes? Does automation retire compulsory effort while preserving or expanding the means by which the person whose effort is no longer required can participate in the common product? Can someone without extraordinary wealth afford the slack necessary to pursue a question whose value is not yet obvious?
This is where the abundance story becomes morally and economically serious. Artificial intelligence is unusually powerful because knowledge is, in important respects, a participable good: my acquiring an insight need not deprive you of it, and an explanation, theorem or piece of code can often be communicated without leaving its source diminished. Digital technology has already made reproduction of many products of intelligence nearly free; AI may radically expand the range of intellectual capability that can be communicated in usable form. But institutions can wrap participable goods in artificial scarcity just as easily as they can widen access to them. A world of extraordinary models behind narrow gates, of vast productivity gains alongside deep insecurity, would represent technological success and social failure at the same time.
True Abundance has to cross over.
The possibility on the other side
Once we put these pieces together, the future of work appears less like a contest to identify the dwindling set of tasks at which humans retain some comparative advantage over machines. That framing has always bothered me because it quietly places the human person in competition with the productive apparatus and then asks him to justify his continued existence by finding something the apparatus still cannot do. It is a strange anthropology: dignity by residual capability.
Lonergan allows us to begin elsewhere. The human being is a subject whose consciousness is dynamically oriented beyond what is already given: experiencing, wondering, understanding, reflecting, judging, deliberating, choosing, acting, and through those acts becoming. Material scarcity constrains that movement without creating it. Indeed, scarcity often prevents questions from being asked because other appetites are simply louder. One has little time for marine biology when the rent is overdue, little room for the violin after two jobs and a commute, little freedom to start the strange little company when health insurance depends upon remaining inside the large one.
What then happens if some of those constraints actually loosen?
We should probably expect plenty of ordinary human vice, foolishness and wasted time. Abundance will not sanctify us. Lonergan, who devoted a great deal of Insight to bias and the flight from understanding, would be the last person to confuse increased capability with authenticity. The same tools that enlarge inquiry can enlarge distraction; cheap creation can produce mountains of rubbish; intelligence can be recruited into egoism, group interest and the endlessly inventive project of avoiding the further pertinent question. More possibility raises the stakes of judgment rather than eliminating them.
But that is precisely why the possibility is worth taking seriously. We have spent most of history forming human lives under conditions in which a large share of practical intelligence had to be directed toward securing material necessities. If technology genuinely changes that ratio, we cannot know in advance what new schemes of human activity might emerge, because the people living inside them will discover goods and practices that our existing institutions have given them little opportunity to explore.
Perhaps many more people become serious amateurs, in the older, richer sense of the word: people doing difficult things from love rather than because a labor market has assigned them the task. Perhaps the forty-five-year-old accountant studies the ecology of a local estuary with instruments and analytical capabilities once confined to a research university; perhaps the retired engineer joins a distributed effort in materials science; perhaps a neighborhood workshop contains machine tools, robotics and biological instrumentation beside the woodworking bench; perhaps three people build a durable firm serving a niche too small to interest today's venture economy but large enough to support excellent work once administrative and technical overhead collapse. Perhaps universities become places people move in and out of over an entire lifetime, less tightly bound to the credentialing of twenty-year-olds. Perhaps we recover forms of civic, artistic, intellectual and religious association crowded out by the demand that nearly every serious adult activity justify itself economically.
And perhaps some of the recovered time is gloriously unproductive. Abundance that merely converts compulsory work into an obligation of ceaseless self-improvement has missed the point. Human flourishing includes contemplation, friendship, worship, play, celebration, care, the meal that takes too long, the book read without any plan to extract content from it, and the walk that records no useful metric at all. We should not automate drudgery only to become more efficient drudges.
This is where I think the Abundance Institute’s historical imagination and Lonergan’s anthropology meet most fruitfully. The former asks us to notice that arrangements which seem permanent can dissolve astonishingly quickly once a constraint is removed; the latter gives us reason to think that the human person revealed afterward will not simply lapse into inertia. Beneath the other appetites there remains the question, and beyond each answer another question, and beyond understanding the demand to judge and decide what is worth doing.
We should therefore be much more ambitious in imagining the prize. The promise of AI is smaller if we conceive it principally as a way to complete today's work with fewer people or in fewer hours. The wild possibility is that cheaper intelligence, joined eventually to abundant energy and increasingly capable tools in the physical world, could widen the range of people able to pursue serious purposes whose pursuit is presently reserved to those with unusual wealth, institutional access or tolerance for precarity. The gain would appear in productivity statistics, certainly, but it might be better perceived in an expansion of agency: more questions followed far enough to become insights, more insights tested against the world, more projects attempted, more people capable of giving concrete form to something they have understood to be worthwhile.
We do not know whether, though I remain rather certain that future will arrive, and the institutions required to make the crossover are not automatic. But we should at least be willing to want it. Our imagination should be large enough to contemplate a society in which fewer people must spend most of their waking lives doing things they would abandon the moment the paycheck stopped, without imagining that the alternative is a population sedated by entertainment and universal basic income. Between drudgery and idleness lies an enormous territory of inquiry, craft, service, enterprise, friendship, art, care, contemplation and adventure that human beings have explored only under conditions of considerable constraint.
The treadmill remains funny because it contains the whole paradox in miniature. We spent centuries reducing the amount of unavoidable exertion required to live and, when we finally succeeded, discovered that exertion itself had not become meaningless. Freed from necessity, we encountered goods within it that necessity had obscured: strength, discipline, play, competition, health, the sheer pleasure of moving through the world under one's own power. The activity returned, transformed by freedom.
Perhaps something like that awaits us with work.
We should want the machines to take the washday. We should want expertise, energy and productive capacity to become abundant enough that the gains actually cross over into ordinary human life. And then, rather than nervously inventing new forms of necessity because we cannot imagine what people might do without the old ones, we might trust the strange creature Lonergan saw so clearly: the creature who, when the noise of the other appetites finally becomes quiet enough, looks out at a reality still overflowing with intelligibility and asks the question again.
What is there to understand?
The really wild possibility is how much more of humanity might finally have the freedom to follow where the answer leads.