This is Part 1 of a three-part series. Part 2: The Reset explores what changes when contribution becomes architecture. Part 3: The Question of Time asks what we do with the hours machines give back.
There is a phrase that has always bothered me: “Tech for Good.”
It sounds noble. It appears on conference agendas and in corporate mission statements, usually next to a photograph of someone holding a tablet in a field. It implies that technology, left to its own devices, drifts toward something other than good — and that a special effort, a special category, is needed to redirect it.
This is exactly backwards.
The phrase creates a distinction that shouldn’t exist. It carves out a small, virtuous corner of the technology landscape and, in doing so, quietly absolves the rest. If “Tech for Good” is a thing, then what is the rest of tech? Tech for indifference? Tech for shareholder returns? The label implies that goodness is an add-on — a feature, not the foundation.
The same logic afflicts Corporate Social Responsibility. When an organisation creates a CSR office, it makes a quiet confession: the rest of the organisation is exempt from the question. Responsibility becomes a department rather than a default, a line item rather than a way of operating. The very existence of the office implies that social responsibility is someone else’s job.
These are not framing problems. They are philosophical errors. And they have a history longer than Silicon Valley.
The Book Everyone Forgets
In 1759, a Scottish moral philosopher published a book that would shape the modern world. His name was Adam Smith. The book was The Theory of Moral Sentiments.
This is not the book people remember.
What people remember — or think they remember — is The Wealth of Nations, published seventeen years later. They remember the invisible hand, the self-interested baker, the free market. They build entire economic ideologies on a selective reading of that later work, often ignoring the foundation Smith had already laid.
The Theory of Moral Sentiments is that foundation. And it says something that would surprise most people who invoke Smith’s name in defence of unfettered markets: the basis of all moral judgement is sympathy. Not self-interest. Not rational calculation. Sympathy — the capacity to feel what another person feels, to imagine yourself in their position.
Smith called this the work of the “impartial spectator” — an internalized observer that asks, before every action: How would this look to someone with no stake in the outcome? The impartial spectator is not a conscience exactly, nor a set of rules. It is a habit of imagination. It is the practice of seeing yourself as others see you, and adjusting your behavior accordingly.
This is not soft philosophy. It is the load-bearing wall of Smith’s entire system. The invisible hand of The Wealth of Nations only works — can only work — because the impartial spectator of The Theory of Moral Sentiments is already at work. Markets function not because people are selfish, but because they operate within a moral framework that makes cooperation possible. Remove the framework and you don’t get a free market. You get a racket.
Smith understood this. His interpreters, for the most part, did not.
The Chartered Confession
Smith was, in fact, deeply skeptical of the corporate form. In Book V of The Wealth of Nations, he argued that joint-stock companies — the ancestors of modern corporations — suffered from a fundamental problem: the people managing the money were not the people who owned it. Directors of such companies, he wrote, being “managers of other people’s money rather than their own, it cannot well be expected that they should watch over it with the same anxious vigilance” as partners in a private firm.
It is worth noting that at least one modern company took this diagnosis seriously. Amazon’s Leadership Principle of Ownership reads: “Leaders are owners. They think long term and don’t sacrifice long-term value for short-term results. They act on behalf of the entire company, beyond just their own team. They never say ’that’s not my job.’” This is, almost precisely, Smith’s prescription for the disease he identified in 1776 — close the gap between action and consequence, and vigilance follows.
This is not an argument against business. It is an argument against the separation of action from consequence — the same separation that “Tech for Good” and CSR offices institutionalize. When responsibility is delegated to a department, the rest of the organisation operates as Smith’s joint-stock directors did: spending other people’s moral capital with less vigilance than they would spend their own.
The modern corporation has refined this separation into an art form. We have Chief Sustainability Officers and ESG reports and impact frameworks, and yet the core business often proceeds as though these functions exist precisely so that it doesn’t have to think about them. The department of good conscience operates so that the departments of revenue do not need one.
Smith would have recognized this immediately. The impartial spectator cannot be outsourced.
The Medical Parallel
There is another tradition that understood this instinctively: medicine.
Primum non nocere — first, do no harm. The phrase is so familiar that its radicalism is easy to miss. It does not say “do good when convenient.” It does not say “establish a department of harm reduction.” It places a constraint at the foundation, before any action is taken. The default is not neutrality. The default is care.
No hospital has a “Medicine for Good” program. The phrase would be absurd — it would imply that the rest of medicine operates without regard for the patient. The Hippocratic tradition embeds responsibility in the practice itself, not in a parallel structure bolted on afterwards.
Technology has no equivalent tradition. Or rather — it did, and forgot. The early internet was built by people who genuinely believed they were building infrastructure for human connection and knowledge sharing. The engineers who designed open protocols and public standards were not doing “Tech for Good.” They were doing tech. The good was the point.
Consider Einar Stefferud — “Stef” to everyone who knew him — who co-designed the MIME protocol that made it possible to send anything other than plain text in an email. I had the privilege of working briefly with Stef in 2000, and what struck me was not his technical brilliance — though it was considerable — but his instinct for generosity. He didn’t patent MIME. He didn’t build a startup around it. He gave it to the world through the IETF, because the point was to make email work for everyone. Or consider Vint Cerf, who after co-inventing TCP/IP — the protocol that is the internet — spent years at NASA’s Jet Propulsion Laboratory designing the Interplanetary Internet: a delay-tolerant networking protocol for communicating across the solar system. A man who had already connected every continent was, in his next act, writing the communications standards for when humanity reaches Mars. That is not “Tech for Good.” That is tech as an expression of what Smith would have recognised as sympathy extended to its furthest horizon — building for people who do not yet exist, in places we have not yet been.
What happened between then and now is not a mystery. What happened is that the separation Smith warned about — between ownership and management, between action and consequence — was codified, theorized, and then industrialized at scale.
In 1919, when Henry Ford tried to cut the price of the Model T and reinvest profits in workers rather than pay special dividends, the Dodge brothers sued. The Michigan Supreme Court ruled that “a business corporation is organized and carried on primarily for the profit of the stockholders.” Legal scholars have since argued that the actual holding was narrow — a question about dividends, not a general theory of corporate purpose — but it hardly mattered. The case became the proof text for shareholder primacy, and culture runs on proof texts, not footnotes. Milton Friedman sharpened the blade in 1970, declaring that the social responsibility of business is to increase its profits. By the 1980s, agency theory had operationalized the doctrine into corporate governance. Venture capital then turbocharged it for technology specifically, creating a class of companies whose primary obligation was to growth, not to the communities they served.
“Move fast and break things” was, charitably, an invitation to challenge bureaucracy and outdated assumptions — to iterate without permission. But a permission structure does not get to choose what it permits. What broke was not only process. What broke was the constraint. Primum non nocere puts a boundary before the first action. “Move fast and break things” removes it, and treats whatever breaks as acceptable cost rather than as the thing you were supposed to prevent.
The Inflection
We are now at a moment where this matters more than it has in decades.
Generative AI is not just another technology. It is a technology that produces language, images, reasoning, and decisions — the things that were previously the exclusive domain of human minds. Its capacity to do good or harm is not incremental. It is architectural. The choices being made now about how these systems are built, deployed, and governed will shape the informational and cognitive infrastructure of the next generation.
This is not the moment for a “AI for Good” program. This is the moment to ask whether the default — the foundation, the load-bearing wall — is contribution or extraction. Whether the impartial spectator is present in the design room, or has been sent to work in the CSR office down the hall.
Smith would say the question answers itself. The impartial spectator is not optional. Sympathy is not a feature. The capacity to imagine the consequences of your actions on others is the minimum condition for participating in economic life.
The technology industry has spent two decades pretending otherwise. Generative AI is the opportunity — perhaps the last comfortable one — to remember what Adam Smith actually said, in the book that everyone forgot.
In Part 2, we ask: what does it look like when contribution is architecture, not afterthought? And in Part 3, we turn to the question Keynes posed nearly a century ago — the question that AI makes urgent again: what do we do with the time?