The Warmer Intelligence

Let me start where the enthusiasts start, because they are right. Artificial intelligence is the most useful thing that has happened to serious work in my lifetime, and I say that as someone who has spent twenty years inside law, capital, and the running of organizations.

What it can do — plainly

It reads everything and forgets nothing, which means the small firm finally has the research desk the big one always had. It drafts — contracts, plans, letters, code — in minutes, so the human hours go to judgment instead of typing. It connects systems that never spoke: the intake form to the calendar to the file to the follow-up, so nobody is lost between them. It sees patterns in the numbers before the numbers become a problem. It translates, summarizes, models, tests, and explains, at any hour, to anyone, patiently.

Deployed well, it does something bigger than all of that. It gives one accountable human the ability to hold a whole — a whole client, a whole company, a whole life — where before you needed a dozen specialists and a prayer that they’d talk to each other. That is the entire premise of the firm I built. I am not a skeptic. I am a user, every day, with my name on the output.

Where it belongs

Alongside the human, never instead of. The rule is simple to say and hard to keep: the machine carries the parts; the human holds the whole. The research, yes. The draft, yes. The analysis, the options, the thousand optimizations — yes, all of it. And then a person with a name and a stake decides, and signs, and answers for it. Every transition I have ever watched succeed — a merger, a succession, a turnaround, a family reorganizing itself around a death — succeeded because a human with authentic values stayed at the center of it. Every one I have watched fail, failed because the center was efficiency, or speed, or a number, and no one was holding the people.

Which brings me to the part that is harder to say, and more important.

We have seen this intelligence before

Here is an uncomfortable observation. The way AI makes decisions — fast, certain, detached, optimizing the stated objective, indifferent to who gets hurt unless the hurt was written into the objective — is not new. It is a very old style of human decision-making, and for most of recorded history it was the dominant one.

Call it what it was. Decisions made from the top, by the few, mostly by men, mostly by men whose seat at the table came from inheritance or force or privilege rather than wisdom. Decisions praised for being unemotional, “rational,” stoic, hard. Empire, conquest, the factory floor, the balance sheet that counted everything except the cost to the humans in it. The style that treated feeling as weakness and empathy as inefficiency. No accountability to the people affected — they were not in the room. No warmth — warmth was not a variable.

That was artificial intelligence before there were machines. A cold, fast, certain optimizer with the power to act and no obligation to feel. We are now building that same temperament into software, at scale, and calling it neutral. It is not neutral. It is the default of the record it learned from, and the record is mostly that history.

AI is not a new kind of mind. It is the oldest kind of power — fast, certain, and unaccountable — finally built into a machine. Which is exactly why the hand that guides it has to be warmer than the history that trained it.

The warmer intelligence

So what do we root it in? Not in a softer version of the same thing. In the qualities the old style dismissed as weakness and that turn out to be the only ones that hold a whole: care for the particular person, not the category. Patience with what cannot be rushed. Generosity that gives to all, not to the few who can pay it back. Accountability that stays in the room after the decision. Compassion as a design requirement, not a footnote.

Our culture has long coded those qualities as feminine, as motherly, and for most of history used that coding to keep them out of the places decisions were made. I am not making a claim about men and women as individuals; I have known men who hold the whole with enormous tenderness and women who optimize as coldly as any machine. I am making a claim about which values get to steer. For thousands of years, the cold ones did. If we let the machine inherit that default unexamined, we will have automated the worst habit of our history and called it progress.

The alternative is deliberate. Prompts that ask who is affected before they ask what is efficient. Systems that measure what happened to the people, not only to the number. Leaders — of any gender — who bring the warmer intelligence into the room on purpose and refuse to apologize for it. That is how you root a transformation in something that lasts: not in the tool, in the values the tool is pointed at.

Two ways to hold a fortune

A contrast from the public record makes the point without a lecture.

Jeff Bezos built one of the most efficient optimizing machines in human history. Amazon is, in a real sense, what the cold intelligence looks like when it works: relentless, data-driven, customer-obsessed, and — by the company’s own famous culture — unsentimental about the humans inside the system. His philanthropy has followed the same logic: large, strategic, branded, directed at problems chosen from the top.

MacKenzie Scott, after the divorce, took her share of that same fortune and did something the old style would never have modeled. She gave it away — well over nineteen billion dollars by her own public accounting — in unrestricted gifts, to thousands of organizations, most of them unglamorous, many of them led by people the big foundations never called. No naming rights. No application process. No strings. Trust placed in the people closest to the problem, and then silence. The essays she published explaining it are among the clearest expressions I have read of the warmer intelligence: generosity as a transfer of agency, not a display of it.

Same money. Same decade. Two intelligences. One optimizes the outcome and keeps the authorship. The other gives the authorship away to the people who will live with the outcome. I know which one I would want programming the machine that will soon help decide who gets what.

What this asks of us

Use the tool — all of it, gratefully. And then insist, every single time, that the human at the center brings what the tool cannot: a stake, a name, and a warmth older than efficiency. That is not a limit on the transformation. It is the only thing that makes a transformation worth having.