What, exactly, would an executive be paying me for? The question has become increasingly important as I have considered a working relationship that would include continuing conversation, strategic advice, careful reading, and help giving language to the moral purposes of an emerging organization. Some of the work could be named in advance: reviewing a document, preparing a brief, shaping a public explanation, or contributing to a particular project. Much of its value, however, would arise before any of those products existed. It would depend upon listening across several conversations, remembering what had already been said, noticing tensions in an unfinished idea, and becoming familiar enough with both the executive and the organization to offer judgment rather than generic advice.

A conventional contract can assign an hourly rate to this work. It can describe deliverables and establish limits. Those provisions are necessary. They protect both parties from the vagueness that sometimes enters a relationship when goodwill is expected to carry more weight than clarity. Yet the contract still leaves something unnamed. The executive would not merely be purchasing a collection of outputs. He would be helping sustain a relationship within which trustworthy attention and responsible judgment could develop.

At almost the same time, I have been thinking through a different question with an artist. Artificial intelligence can now generate an image from a prompt, revise an existing conception, and imitate many of the visible signs of artistic labour. It can produce an attractive object quickly. Copyright law can help determine who owns an original image, what permissions have been granted, and how a later image may legally be used. But the legal questions do not quite reach the moral centre of the matter. What relationship should remain between an artist and work that has grown from the artist’s imagination, skill, and earlier labour? When an AI rendition stands between the original work and a new use, who is still present in what has been made?

Together, these questions returned me to an argument I made during the first months of the pandemic about the value of work—and showed me where that argument remained incomplete.

In “Value of Work,” I argued that the market price of labour cannot adequately measure its contribution. The pandemic made this unusually visible. During the pandemic, workers whose labour sustained everyday life received forms of public recognition that their wages had rarely expressed. Their work did more than generate income or contribute to a company’s profitability. It sustained biological life and helped preserve the social world in which life could continue.

I called this the constitutive value of work. Work helps build the durable world in which human life can appear and endure—not only its physical infrastructure, but also its institutions, practices, and relationships. Hannah Arendt called this the human artifice. Its value is easily obscured when work is measured only by salary, productivity, profit, or consumer demand.

The language of economics sometimes calls these contributions positive externalities. A teacher may be paid to deliver a course, while the course also restores a student’s confidence and enlarges her sense of possibility. The price of the service does not capture the full benefit created through it.

That argument remains important, but “externality” is now beginning to seem too weak. It assumes that the transaction or product sits at the centre while trust, recognition, encouragement, belonging, and moral attention gather around its edges. In many forms of work, those relationships are not incidental benefits. They are part of the substance of the work itself.

What economic language calls an externality may sometimes be the very thing people entered the relationship to receive.

This is evident in the possible relationship with the executive. He would not be paying for friendship, and friendship should never become a pleasant word used to hide the expectation of unpaid labour. Nor would a financial agreement give him ownership of my judgment or unlimited access to my time. A healthy arrangement requires boundaries, specified responsibilities, and honest compensation. Yet it can still be covenantal in a modest sense: each person commits to protect the conditions under which trust, candour, continuity, and mutual benefit can grow.

The executive may need a document at a particular moment. It will be better because accumulated understanding may allow me to recognize that a seemingly harmless phrase conflicts with the organization’s earlier commitments. Such judgment cannot be fully specified before the relationship begins. It is formed through attention, dialogue, and repeated acts of reliability.

This kind of work is often misdescribed as the sale of expertise. Expertise matters, but expertise detached from relationship is readily exchanged for another expert. Relational judgment develops when knowledge becomes responsive to particular people, histories, purposes, and circumstances. The person offering it is not merely applying a technique. He is receiving a situation, interpreting what is being asked, and deciding how to answer. Payment makes that attention sustainable. It does not manufacture its value, and it should not pretend to own the relationship from which the value comes.

The artist’s situation exposes another limit of a product-centred account of work. A finished image can be copied, licensed, purchased, altered, uploaded, or fed into an AI system. Once it circulates as a file, it becomes easy to treat the image as though it had separated cleanly from its origin. Yet the work carries the artist’s learned habits of attention, experiments, discarded versions, technical choices, memories, and interpretations. It also carries the relationship between the artist and the person who commissioned or received it.

Copyright is necessary because creative people need enforceable rights. It tells others that an image is not simply available because it can be copied. It can establish permission, attribution, ownership, and compensation. Still, copyright gives us only a partial moral vocabulary. A person can possess a legal right and exercise it poorly. A later image can avoid a narrow infringement test while continuing to depend upon the recognizable labour of someone whose contribution has been made invisible. Conversely, a creative collaboration can be generous and truthful even when several people, traditions, references, and tools have contributed to the result.

Artificial intelligence intensifies the difficulty because it can preserve the appearance of labour while obscuring the person who laboured. A generated image may display polish, composition, symbolic coherence, and a recognizable visual language without showing who learned to see in that way, who accepted the risks of experimentation, or who can answer for the final choices. The machine can give us something that looks finished before a human relationship has decided what completion should mean.

I have made a related argument about writing in “Authorship, AI, and the Responsibility to Answer for What We Write.” Authorship is more than typing every word alone. Writers are helped by books, teachers, editors, conversations, traditions, memories, and tools. What preserves authorship is the writer’s willingness and ability to answer for the result: to explain it, defend it, revise it, identify its sources, and take responsibility for its errors. Assistance can deepen authorship when it passes through human judgment. It hollows authorship out when it allows the appearance of understanding to replace understanding itself.

The same standard can be extended to visual art and perhaps to work more generally. The decisive question is not whether a tool entered the process, but whether the human participants remain present and answerable within the result. In this case, that means asking whether the artist consented, whether the contribution is truthfully recognized, and whether AI supported or bypassed the artist’s judgment.

In my own situation, the more responsible path is not simply to accept the generated rendition as a cheap replacement for the artist’s work. It is to let the generated conception become part of a renewed human conversation. The artist can interpret it, reject it, transform it, and produce something genuinely new through his own judgment. The resulting work may still carry the history of the earlier image and the AI-assisted experiment, but those influences no longer need to erase the artist. They can be placed under an authorship for which a person is prepared to answer.

To stand behind our work is to remain answerable for what we have made. But the responsibility does not belong only to the maker. Those who receive and benefit from work should also recognize the people and relationships that made it possible. Authorship and solidarity meet here.

This fuller account also changes how I understand the book I have been writing. A manuscript becomes a durable object: it gathers thoughts that might otherwise remain dispersed through memory and conversation. In Arendt’s sense, it is clearly work. Yet the value of the manuscript cannot be contained in the object.

As I wrote in “Writing Without a Guarantee,” I cannot control the book’s reception. I can control whether I return to the desk and submit each claim to the disciplines of evidence, revision, and criticism. These acts do not guarantee an audience. They are how I become answerable for the offering.

The book has also formed me while I have formed it. Ideas that began in philosophy have been tested against experience, and experiences that once seemed disconnected have gradually acquired language and relation. The writing has exposed where abstraction concealed me and where greater truthfulness was required before the sentence could become clear.

This does not mean that everyone should seek an identity through employment, or that every demanding project is secretly a vocation. Work can deform people. It can exploit devotion, consume attention needed elsewhere, and make recognition conditional upon performance. Meaningful work can become a flattering excuse for neglect. A worker may need to leave a job precisely because the institution has separated the contribution from the person and now wants the product without the voice, limits, or continuing presence of the one who made it.

The possibility of formation nevertheless matters. Work can disclose capacities that private self-invention could not reveal. A teacher may discover that she can call hesitant students into speech; a writer may discover an argument only by submitting to the discipline of the page. We do not simply express a finished self through work. We become more capable of answering through it.

Here the language of gift matters. In “Fulfilled Needs and Desires: by Gift or by Effort?” I resisted the fantasy of the self-made person. Our capacities arrive through bodies, communities, traditions, and acts of recognition that we did not create for ourselves. Effort develops what was first received.

Work becomes one way of caring for what we have received. We develop inherited capacities and return them, altered by attention and judgment, to the common world. None of us begins from nothing, and none of us can truthfully claim complete ownership of the conditions that made our contribution possible.

This makes responsibility larger rather than smaller. To call a capacity a gift does not release us from answering for its use. It means that our work is never merely private property or autonomous achievement. What we make enters relationships already in progress. It draws upon a world we have received and alters the world others will inherit. We should therefore be able to say more than “I was paid,” “I completed the task,” or “I own the copyright.” We should be able to say how the work honours the people who contributed, what purposes it serves, and why we are prepared to place it before others.

A A fuller valuation of work must therefore ask about compensation, the thing made, the persons and institutions formed through making it, and the person who remains answerable. Each question catches something the others miss. Fair pay cannot substitute for recognition, and recognition cannot substitute for pay. Outputs must also be judged by what they do to workers and communities, while meaningful work still requires someone willing to answer for its consequences.

Compensation without recognition treats the worker as an input. Recognition without compensation praises people while asking them to subsidize everyone else’s good intentions. Production without constitution counts outputs while ignoring what those outputs do to workers and communities. Constitution without answerability can romanticize meaningful work while leaving no one responsible for its consequences. A just account must hold all four together.

This is especially urgent as organizations adopt artificial intelligence. The appeal of AI is often presented as efficiency: more products, produced more quickly, at lower cost. Efficiency can free people from repetitive tasks and make valuable capacities more widely available. It can also encourage organizations to capture the accumulated language, images, judgment, and styles of workers while gradually removing those workers from the process. The deepest danger is not merely that a machine will make a better product. It is that institutions will decide the product is all they ever valued.

The same danger existed before AI. Institutions have always been tempted to preserve the usable results of a person’s contribution while treating the contributor as expendable. AI gives that old temptation a powerful new instrument: it can preserve the surface of a contribution after the person whose judgment formed it has been made absent.

The response cannot be a romantic demand that every product remain forever under the control of its maker. Work is meant to enter a shared world and take on uses its maker cannot fully govern. Release is part of work’s dignity. Arendt understood that once words and deeds enter the web of human relationships, their consequences cannot be mastered by the person who began them.

Release, however, need not require erasure. A work can pass into other hands without denying its history or severing recognition from the person who made it. Consent, attribution, compensation, and continuing answerability allow work to become shared without pretending that it came from nowhere.

Good work leaves a person present within the world that person has helped to make, even after the product has begun a life beyond its maker.

This extends my earlier account of constitutive work. Work does more than sustain life and build a human world; it binds people morally to what they make and to those who receive it. Those bonds can be honoured or exploited, and they belong to the value of the work itself. The practical implications are modest but real. Formal agreements matter because they can protect relationships that the agreements themselves cannot create. A contract, a copyright agreement, or a publication plan should make responsible work sustainable without reducing the worker to the product.

Perhaps this is what it means to stand behind our work. We accept the labour, assistance, inheritance, judgment, and risk that have entered it. We acknowledge what we have received, accept responsibility for what we make, and refuse to let a contract or a machine answer in our place. And when we receive the work of another person, we resist the temptation to take the product while allowing the worker to disappear.

Beyond the product is a person who has attended, learned, risked, judged, and answered. Beyond the transaction is a relationship that can make such attention possible. Beyond ownership is a common world in which what we make passes into other hands and begins to mean more than we can control. The value of work is found in all of these places. A humane economy and a truthful use of technology will depend upon our willingness to keep them connected.