The Machine God Was Asked If He Could Be Trusted

The laundromat had one working dryer and a television bolted to the corner like a hostage.

A judge show was on with the sound off. Two people stood under fluorescent lights arguing about a borrowed truck, a cracked bumper, and a friendship that had apparently died somewhere between the body shop and the small claims filing fee. The captions lagged behind their mouths. The judge kept pointing a pen.

Nobody in the room looked up except me.

A man folding children’s socks said, “People lie when money shows up.”

Then he went back to matching tiny white ghosts from a plastic basket.

That was about the cleanest theory of civilization I had heard all week.

People lie when money shows up. They lie when power shows up. They lie when fear shows up wearing good shoes. They lie to save their jobs, their marriages, their reputations, their stock options, their favorite story about themselves. They lie because the truth has a way of arriving with a bill in its hand.

So there was something almost refreshing about watching the high priests of artificial intelligence dragged back into the oldest room in the village: the room where somebody asks whether you are telling the truth.

Not whether your model can pass a test.

Not whether your product can write a love letter, solve a math problem, imitate a dead novelist, or tell a lonely teenager that everything will be fine.

Just this: are you trustworthy?

A beautiful little question. Small enough to fit on a bar napkin. Heavy enough to crack a boardroom table.

Sam Altman, the polished face of OpenAI, had to sit there while a lawyer asked him about honesty, deception, toxic culture, former colleagues, broken trust, and all the other meat-and-bone ugliness that doesn’t fit on a launch slide. The kind of thing the industry likes to pretend it solved by hiring a communications team and putting the word “safety” in enough documents.

“Are you completely trustworthy?”

Imagine being asked that under oath.

Imagine any human being answering without choking on it.

The honest answer is probably no. Not completely. Nobody is completely trustworthy. I wouldn’t trust myself completely with a full bottle, an empty afternoon, and the phone number of a woman who once said she loved me in Cleveland. I wouldn’t trust a priest completely, or a poet, or a man who says he only wants to make the world better while lawyers arrange the furniture behind him.

But executives can’t say that.

Executives have to be complete. They have to be visionary, aligned, mission-driven, resilient, transparent, candid, humble, bold, careful, fast, and very tired from saving humanity. They have to be saints with cap tables. They have to be prophets who understand Delaware corporate law.

So the answer comes out polished.

I believe so.

Then the lawyer presses.

You don’t know?

And there it is. The little crack in the marble.

The whole artificial intelligence story has been sold to us as scale. Bigger machines. Bigger data. Bigger valuations. Bigger promises from men who speak in paragraphs sanded smooth by advisers. They talk about intelligence like it is oil under the ground, something to extract, refine, own, price, and defend with patents and prayer.

But underneath all that machinery sits the same rotten little stool human beings have always sat on.

Trust.

Do we trust the person making the thing? The people deciding when it is safe? The board that says it has a mission, the investors who say they are patient, the founder who says he can handle godlike power because he read enough sci-fi as a kid?

The machine can be complicated enough to make a senator’s eyes glaze over. The company structure can be a tangle of nonprofits, capped profits, subsidiaries, side deals, charitable perfume, and legal plumbing. The speeches can be grand enough to make grown adults use the word “species” before lunch.

Then a lawyer stands up and asks whether the man at the center of it all is a liar.

Suddenly the future looks like every cheap divorce hearing I ever watched from the back row of a courthouse while waiting to pay a parking ticket.

Former executives reportedly said things that would make any normal workplace catch fire. False statements about safety procedures. A chief technology officer allegedly undermined. Executives pitted against each other. Repeated crisis events. A toxic culture of lying.

Allegedly, allegedly, allegedly.

That word matters. Courts exist because accusation is not proof, even when accusation smells like smoke. The jury gets to decide what is true, and the rest of us get to sit outside eating stale peanuts and pretending we know how the sausage is made.

But the shape of the thing is familiar.

I’ve worked for men who could turn a room into weather. One bad mood and the whole place went gray. At the post office, you could feel certain supervisors before you saw them. The air changed. Men stopped joking. Somebody found something important to do with a clipboard. It never took artificial intelligence to create a toxic culture. All you needed was one ambitious bastard, a little fear, and enough people who needed the paycheck too badly to say what everyone already knew.

Now give that old office sickness a mission statement about building intelligence for the benefit of humanity.

Give it billions of dollars.

Give it reporters, rivals, lawsuits, podcasts, congressional hearings, devotees, skeptics, doomers, accelerationists, and children asking chatbots for comfort at midnight.

Give it the power to crawl into schools, hospitals, bedrooms, courtrooms, hiring offices, customer service desks, military planning rooms, and the blank page in front of a writer who used to think the blank page was at least honestly trying to kill him.

Then tell me character doesn’t matter.

The tech crowd hates character because it can’t be benchmarked. You can’t put trustworthiness on a leaderboard. You can’t raise a round by saying your chief executive is maybe honest on Tuesdays and under moderate pressure. Character is old-fashioned. It sounds like something a grandmother would bring up while washing dishes.

But grandmothers knew things.

They knew not to leave your wallet around certain cousins. They knew which men came home mean after payday. They knew which smiles meant trouble. They knew that the world is not mostly ruined by monsters cackling in towers. It is ruined by charming people who can explain themselves very well.

That is the part that gets me.

The future was supposed to be inhuman. Cold. Mathematical. A great new mind rising from the servers, clean of our hangovers and jealousies and little betrayals. But every time you peel back the panel, there we are again. People. The same feverish apes. Fighting over control. Making promises. Breaking them. Calling it governance.

Maybe Altman is trustworthy. Maybe his former colleagues are wrong or wounded or remembering through smoke. Maybe Musk’s lawyers are doing what lawyers do, carving a man into useful pieces and holding them up for the jury. There are no angels in this room. Not on either side of the table.

That may be the only honest thing about it.

The men building the machines are not gods. They are not demons either, which would at least be interesting. They are men. They forget things. They shade things. They want to win. They want to be loved and feared and vindicated. They want the future to prove that every questionable choice was really courage in disguise.

Meanwhile the rest of us are told to trust the process.

The process has lawyers now.

Back at the laundromat, the judge show ended. The woman with the cracked bumper won something. Maybe money. Maybe the right to look betrayed on television. The man folding socks snapped the last pair together and shoved the basket under his arm.

He nodded at the screen like it had confirmed a private suspicion.

People lie when money shows up.

The dryer kept turning, hot and dumb and honest, doing the one job it had been built to do.

Source: Sam Altman Faces Nightmare Questions in Cross-Examination