The Students Are Right

This spring, at commencement ceremonies across the country, the mention of artificial intelligence was met with booing.¹ Not heckling a politician. Not polite disagreement. Booing. From graduates. At their own ceremonies, in the middle of the speeches meant to send them into the world.

I work in the industry being booed at.² I want to talk about it plainly, because plainness is what the moment asks for, and plainness is the one thing my industry has decided it cannot afford.

Start here: the students are right to boo.

I say that first because most of what has been written since came from people whose opening move was to explain the students to themselves. They don’t understand the technology. They’ve been spooked by alarmists. They’ll come around once the productivity shows up.³ This is the sound an industry makes when it has decided that the people most exposed to its product are the people least entitled to feel anything about it. It is the voice you use when you are hiding something and hoping nobody watches your hands.

The students are watching the hands. They are twenty-two, not stupid. They spent four years and a six-figure tuition preparing for a job market where recent graduates are now unemployed at a higher rate than the general workforce — the fifth straight year that has been true — and where more than four in ten of the ones who do find work are underemployed, holding down jobs that never needed the degree they are still paying for.⁴ They are right to be afraid. They are right to be angry. And they are right that most of the people onstage promising them it will be fine are the same people positioned to do fine regardless.⁵

Under the circumstances, booing is a restrained response.

The problem with my industry is not the technology. The technology is what it is. Some of it will be a gift. Some of it will be a wreck. Most of it will sit in the long ambiguous middle we will not understand for ten years. That is true of every technology. That is not the scandal.

The scandal is the vagueness.

Every public sentence about what this does to working people runs through a filter built to make it impossible to act on. Augmenting human capability. Reshaping the nature of work. Unlocking opportunities we can’t yet imagine.⁶ Each phrase does the same job: it takes a question a person could answer and swaps in a weather system nobody can argue with. The question is will my job exist in five years. The fog is the future of work is being transformed.

I have been in the rooms where the honest answer was sitting right there on the table. I have watched it get sanded down to fog before it left the building. I have done some of the sanding myself.⁷ I am not the villain here, and neither are my colleagues — most of them are decent people doing what looks, from the inside, like reasonable work. But a thousand decent people each smoothing their own small edge produces a public conversation in which the people who know the most hand the people who need it most the least useful version of the truth.

The students are booing the fog. Good.

Here is what the fog is built to hide.

There are two different technologies wearing the same two letters, and the marketing needs you to keep confusing them, because the confusion is the whole business model.

The first one is an exoskeleton. It straps onto people who are already capable and makes them faster. The lawyer who researches in an hour what used to take a week. The engineer who ships on Friday what used to eat the month. This is the version in every keynote, and it is real, and it carries a cost the keynotes skip: an exoskeleton has no use for the person on the bottom rung. The junior analyst, the first-year associate, the entry-level anything — those were never quite jobs. They were apprenticeships. The years you spend being not-yet-good in exchange for becoming good. The exoskeleton eats the apprenticeship. That is the thing the graduates can smell from their seats. They are not afraid of the future. They are afraid that the bottom rungs of the ladder were sawed off while they were climbing toward them.

The second technology is a prosthetic. A prosthetic does not make a capable person faster. It restores function to someone who could not participate at all. That is not productivity. That is access. The single mother who cannot take an unpaid day to sit in a county office gets the same guidance as the man who can pay someone to handle it for him. The kid whose benefits letter is written at a fourteenth-grade reading level finally has it read back to him in a language someone actually speaks. The prosthetic does not replace the relationship between a person and the system that is supposed to serve them. It makes that relationship possible for the people it was never available to.⁸

Same two letters. Opposite moral direction. One compounds the advantage of people who already have it. The other hands a floor to people who have been falling through it their whole lives.

I care about the exoskeleton because it pays my mortgage. I care about the prosthetic because it is the only version of this I think is worth a damn. And the reason the booing lands is that nobody onstage is drawing the line between them. They are selling the exoskeleton and dressing it in the prosthetic’s clothes — borrowing the moral credit of access for the disenfranchised to move a product whose actual near-term effect is to erase the disenfranchised’s first job. The students do not have that vocabulary yet. But they can hear that the two things have been welded together on purpose. They are booing the weld.

I am not going to tell you what to do, graduate. I do not know what you should do, and anyone who says they do is selling or lying.⁹ The next five years are going to be strange, and nobody — not the people building this, not the people regulating it, not the people writing essays like this one — knows the shape of the strangeness yet.

But I will give you the one thing I have asked of myself.

Assume the part of your education that made you a competent processor of information is worth less than it was the day you enrolled. Assume the part that made you a person is worth more. Judgment. Taste. The nerve to know a thing is wrong while the whole room insists it is fine. The capacity to care about one specific human being and do something about it. The machines are good at the first and bad at the second, and the evidence that they are about to get good at the second is thinner than the people selling it need you to believe.

Do not curdle into nihilism, because nihilism is the outcome the people in charge are quietly rooting for. A demoralized graduating class does not organize, does not vote, does not make demands, does not notice when its government stops returning its calls. The fog has an interest in your not noticing. So notice. Loudly. You have already started.

And do not let anyone — me included — talk you out of being scared. The fear is accurate. In this case it is a better read of the situation than the optimism being handed down from the podium. Trust it. Then do what people have always done with fear once they stop being paralyzed by it: use it to work out the next move, slowly, in the company of people who love you.

My industry has a thing it tells itself at two in the morning, when the question of what it is actually doing comes around. The thing goes: someone is going to build this. If the people who care do not build it, the people who do not care will. So the people who care have to be in the room.

I have made that argument. I have heard it made. It is not stupid. It is also the exact argument every powerful institution in history has made for its own continuation, and the record on whether the people in the room keep caring once the room gets comfortable is mixed at best.¹⁰ The argument does not validate itself. It has to be re-earned, constantly, by what the people in the room actually do.

So here is what I would ask of the people in the room — my colleagues, the people above me, the founders whose names are on the buildings. Stop hiding in the fog. It was never protecting the public. It protects us. It lets us keep our salaries and our self-image, lets us call ourselves the good guys on the grounds that we personally have not fired anyone. The fog is the moral hazard. The fog is the thing being booed.

It can be done, because it has been done. A year ago the chief executive of one of the leading AI labs sat down with a reporter and said the plain version out loud: that this technology could erase half of entry-level white-collar jobs inside five years, that unemployment could climb to ten or twenty percent, that — his words — “we, as the producers of this technology, have a duty and an obligation to be honest about what is coming.”¹¹ He said it the same day he stood onstage selling his own models. He is not a doomer and he did not quit. He simply refused to sugarcoat it. And the most instructive part was not the warning. It was the reaction. Much of the industry flinched, called it fearmongering, and went straight back to unlocking opportunities we can’t yet imagine. When one of the people who knows the most tells the truth, the fog does not lift. The fog closes over him.

Say it anyway. Tell people what is coming. Tell them what you do not know. Tell them where this helps and where it harms and which of the two you personally worked on this week. If you cannot say it in public because it would cost you your job, say it in private, to the people in your life who need to hear it. Tell your nieces and nephews. Tell the kid from your hometown about to sign loans for a degree in a field that may not be there. Tell your own children. Tell them what you actually think.

The graduates are not asking us to stop building. They are asking us to stop lying. That is a smaller request than the panic-readers are making it sound. It is also a far more important one.

I am writing this from inside. I am not quitting. I am not staging a conversion. I have a family to take care of, and I believe, on the balance, that the work I am doing in here is more useful than the work I would do out there. I made those choices and I will answer for them.

What I will not do is pretend the kids are wrong. They are not. When I look honestly at the thing I help build, I do not see a god and I do not see a monster. I see a mirror — a surface that shows us what we already are, the cruelty and the care both, and then hands us the bill for the one we decide to scale. The doom we keep bracing for is our own reflection. So is the way out.

Nobody is coming to settle this for us. Not a superintelligence, not a regulator, not the market. We are the ones in the room, and we are all we have got — which is not the tragedy it sounds like. It is the starting point. It means the fog is a choice, the prosthetic is a choice, the honesty is a choice, and the choosing is still ours.

The students saw all of that and named it in the only language they had on a Saturday morning under a tent, which was booing. They were right. They are owed a better answer than condescension. They are owed the truth, plainly, while it can still do some good.