The Human Core of Value

Anyone who has advised people for long enough has taken a call like this one. A client phones late at night about a deal. You talk about the deal for four minutes. Then you talk for most of an hour about his business partner, his marriage, his fear that he has built something he no longer recognizes and whether he is the kind of person who could walk away from it. The deal was the reason for the call. It was not the reason he called you.

Anyone who has spent a career solving human problems for other people knows that call. The business advisor finds herself acting as priest, rabbi, therapist and confessor. The lawyer becomes, depending on the hour, a bartender, a sparring partner, a confidant, a friend and sometimes something close to a parent. The accountant hears about the divorce before the spouse does. The banker is the one person the founder can admit fear to. None of these roles appears on the engagement letter. All of them, in my experience, are where much of the real value lives.

An old friend and colleague put it to me once in a way I haven’t been able to shake. A piece of software, he said, can make a client feel comfortable. It will never make them feel safe. Safety comes from a person: someone with judgment, who is accountable, who has been in this situation before and who will still be there when it goes wrong.

The task and the role

I’ve come to think there is a distinction we rarely make explicit, between the task and the role.

The task is what is written down: draft the agreement, model the scenarios, prepare the board deck, review the code, negotiate the terms. It can be specified, scoped, priced and billed. For most of the history of professional work, the task has also been the unit of account. We bill for it, staff for it, train for it and measure it.

The role is everything else. It is the trust that lets a client tell you the real problem instead of the presenting one. It is the judgment to know which of five true things matters this week. It is presence in a hard room and the courage to say what nobody else will. It is care. The role is why a client calls you at eleven at night about something that has nothing to do with the contract.

For a long time the two were bundled so tightly that we could pretend the task was the product and the role was just good service. Nobody had to decide which one they were really paying for.

The bundle is coming apart

That is changing quickly. Machines are becoming extraordinarily good at the task. Give a modern AI system a well-defined problem with clear boundaries and it will draft, summarize, analyze and model in seconds what used to take a team a week. Much of what professionals were nominally paid for is getting cheap and in some cases nearly free.

There are two ways to read this. One is as a story of replacement: the machines are coming for the professions and what is left is a race to the bottom. The other is that the machines are doing us a strange favor. By taking over the parts of the work that can be written down, they are revealing the parts that can’t. They are pulling the role out from behind the task and showing us, perhaps for the first time with discomforting clarity, what we were really paying each other for all along.

History offers some comfort and some caution. When the printing press arrived, scribes did not simply disappear. Value moved toward editing, judgment, curation and trust and cheaper books created far more demand for reading and writing than anyone expected. But the people who had defined themselves by the act of copying had a hard time and the transition was neither quick nor gentle. Every large technological shift, from the clock to the factory to the internet, has changed how people relate to their work and to each other. What has persisted through each one is the thing underneath: people needing other people they can trust.

The question I want to ask

Most of the conversation about AI right now is about what it can do. That is a good question and plenty of capable people are working on it. I want to ask the inverse. Assume the machines will keep getting better at every well-defined task. What is left? What, in every domain, can they not do? Where are the human gaps and what are they worth?

But this is just the obvious first question. The more meaningful questions are what are these things called “trust” and constituent inputs of trust such as “judgment” and “accountability”? Might there be many ways to think about each? And might those different understandings give us very different pictures of the way humans interact, share resources, live together and build economic and social structures? What do we mean when we say these things have worth? What has caused these things to be filled out with some meaning that we relate to and collectively understand, in one context or another, as foundations of value in economic life?

To begin a dialogue with more granularity, these three words cover several mechanisms with different origins and very different economic fates. The most useful theme in the literature is that major technologies don't eliminate the need for trust. They move trust from one mode to another and value collects in whichever modes the new system can't produce. There are a few threads in this theme that are worth exploring. I outline them here as a roadmap for further inquiry:

Risk and its two faces. Trust, judgment and accountability each take two forms. Trust can rest on assurance: the security of closed relationships in which betrayal is costly. Or it can rest on a leap, extending reliance where betrayal is possible, which opens people to new partners and opportunity. Judgment can be prudent, guarding against loss. Or it can be bold, accepting error in pursuit of gain. Accountability can mean answering for following the procedure or answering for the outcome. Both forms create value, yet cultures and minds weight them very differently, and each carries its own irrationalities. On the cautious side, blame falls harder on action than inaction, losses loom larger than gains, decisions bend toward what can be defended rather than what is best and novelty praised in principle is rejected in practice. On the bold side, confidence outruns evidence, control is overestimated and the costs of failure are underweighted or shifted onto others. Groups welcome goals reached by unapproved means while condemning the means and treat what fits no category as dangerous whatever its results. As machines take over tasks and even learn to detect these patterns, managing them remains human work: recognizing which irrationality is at play, in whom, and bringing enough order to it that people can take productive action. These things require standing, trust and accountability that a machine does not hold. The open questions are how each culture, institution and person weights the risk-averse and risk-taking forms; why that weighting so often diverges from the value each produces; whether welcoming the fruits of risk while recoiling from its manner is incoherent or a consistent logic of order and blame; and what it takes to bring order to the irrationalities on either side.

Technology converts personal trust into system trust. Sociological research on American industrialization from 1840 to 1920 shows trust based on kinship, religion, custom, ethnicity, obligation and reciprocity giving way to institutional trust: credentials, contracts, regulation, insurance. Such "abstract systems" still need human access points, such as the doctor or banker through whom people meet the system face to face. Their value rests on several distinct grounds. Some come from the system itself: fluency in how it works, credentials it confers, and control over who gets in. Some come from standing between the system and the person: discretion to apply rules to a particular case, translation in both directions, answering for failures, and repairing trust when things go wrong. Some come from outside the system entirely: character, reputation and relationships it does not control. Across all of these runs facework, the composure and attentiveness that reassure people the system is in good hands. AI is the next abstract system. The open questions begin with facework: whether reassurance can be performed by machines, and whether it still reassures once people know it is performed. Beyond that: which grounds of an access point's value erode as systems explain and certify themselves, which grow as systems become more capable and opaque, whether the access point remains the system's agent or becomes the person's agent in dealing with it, and what becomes of access points when people face not one system but many.

The asymmetry of stakes. When you depend on someone, you can be reassured in two ways: by evidence that they will perform reliably or by the fact that they stand to lose something if they fail you. Machines increasingly provide the first, and where proven reliability is enough, they will replace human trustworthiness. But a machine cannot provide the second. It has nothing to lose and no interests of its own that it sets aside on your behalf. Thus, its performance can be dependable but never loyal. Technology therefore separates bearing the consequences of work from doing the work. Because reliable systems make failure cheap to insure, even routine accountability may become a commodity. The open question is which forms of exposure resist insurance, and how economic life reorganizes around the people who carry them.

Cheap prediction, valuable judgment. Economists have divided decisions into prediction (what is likely to happen) and judgment (how much each outcome matters). As AI makes prediction cheap, demand for judgment rises, as every prediction still needs someone to weigh the outcomes. But much of what passes for expert judgment is really prediction: the seasoned professional's intuition is pattern recognition and it is genuine only where stable patterns and rapid feedback let it be learned. These are the same conditions machines need. So machines will absorb skilled intuition along with prediction. What remains human is judgment where the weight of each outcome is unclear, contested among stakeholders or being set for the first time. The open question is whether this work consists of uncovering weights people already hold but can't articulate, or helping them form weights they don't yet have, and how different those two practices are.

There is a lot of nuance to unpack, but it is clear that the lasting source of value between people has always been relational: trust, judgment, presence, accountability, the feeling of being understood and the confidence of being protected. Anthropologists have articulated for a century that exchange is never only about the thing exchanged. It is also about the relationship it creates. Economists are starting to describe a “feeling economy” in which, as analysis gets cheap, human connection and empathy become the scarce goods. And anyone who has sat across a table from a frightened client already knows it.

An invitation

This is the first of a series of pieces I’ll be writing over the coming year about where value really comes from in economic life. Part of that will be about how artificial intelligence is bringing that question into focus. But, that is only a question that begs more questions. And artificial intelligence itself tells us nothing about either the reasons we ask the question or even the beginning points of answers. That is the domain of psychology, anthropology, sociology, economics and mysticism.

I don’t have all the answers and I’m wary of anyone who claims to. So I’d like this to be a conversation rather than a lecture.

If you do work that solves human problems (medicine, teaching, ministry, law, advising, finance, design, care), I’d like to know:

When a client trusts you, is it because they've seen you deliver before, or because they know you have something to lose if you fail them?

When you've taken a real risk on a client's behalf, how was it received, both when it worked and when it didn't?

How much of your work is helping people discover what they already want and how much is helping them decide what they should want?

When clients disagree among themselves about what matters most, what do you actually do?

Which of your judgments could someone learn from a manual or enough repetition, and which could they not?

What is the most irrational thing you regularly help clients get past, and how do you do it?

What do clients need from you that the institutions you work within (the firm, the hospital, the court, the bank) can't give them?

When you represent a system to a client, whose side are you on, and has that changed over your career?

How do you show a frightened person that things are under control when you aren't sure they are?

What have you seen that rebuilds trust after it has been broken?

Write to me. The most interesting answers will shape what comes next.