The Doomsday Nobody Discusses

In December of 2023, the Senate convened a forum to talk about the end of the world.

That is not quite how the invitation read, but it’s close. The title was Risk, Alignment, & Guarding Against Doomsday Scenarios. Leader Schumer had assembled a run of these sessions to figure out what to do about AI, and Senators Rounds, Heinrich, and Young were in the room. I was asked to submit a written statement and testify. On the biology side, the doomsday everyone had come to guard against was the familiar one: a person with no training and bad intent sits down at a laptop, asks an AI the right questions, and walks out with a pandemic.

I’d spent years by then inside the machinery that’s supposed to steer this — the Pentagon, a national security commission, the labs. So I understood the assignment. Name the catastrophe, then guard against it.

The most useful thing I could do in a room built to catalog doomsdays was to name the one that wasn’t on the list.

Not the pandemic in the garage. The other one. The scenario where fear itself does the damage — where we grow so frightened of what a technology might do that we restrict it into uselessness, and people die anyway. Not from a pathogen. From a moratorium. People die because a limitation hindered the drug that would have saved them. People go hungry because our ability to advance the food supply stalled. That is also a doomsday. It just doesn’t look like one, because catastrophe by omission never photographs well.

I did not have a formula to hand them, and I said so. You cannot prove an absence of risk. What you can do is build empirical ways to measure how it changes over time. The question that actually matters is narrow: how much does AI increase risk compared to a baseline? Without a baseline, you can’t measure the thing you claim to be afraid of. The room wanted a verdict — safe or not safe, allow it or ban it. I offered instruments instead, because instruments are the honest thing to offer when the truth is that nobody knows yet.

That could sound like a one-hearing contrarian streak. It wasn’t. It was the compressed version of an argument I later wrote out at full length, and the long version has a name I borrowed on purpose.

Naomi Klein called it the shock doctrine: the strategy of using a crisis, or the fear of one, to push through extraordinary measures that would never survive an ordinary debate. Klein was writing about disaster capitalism — shock used to force deregulation, privatization, the grab for power while everyone’s too rattled to object. What I kept noticing is that the same machinery runs straight through the life sciences, with the payload reversed. Here the shock is a vivid biological nightmare, and it’s used to push the opposite direction: freeze the research, ban the tool, wait until we’re sure.

The pattern is old and it repeats almost on schedule. Recombinant DNA at Asilomar in 1975. “Frankenfoods.” Venter’s synthetic cell and the “playing God” headlines. The 2022 experiment where a drug-discovery AI, run in reverse to maximize harm instead of minimize it, spat out forty thousand toxic molecules in six hours and became, in the retelling, proof that anyone could now print a bioweapon. Mirror life, most recently, with a serious call to halt an experiment nobody has actually run. Each time the fear was loud and specific, and the response was restriction.

And each time — this is the part the nightmare skips — the predicted catastrophe didn’t arrive, and the benefits did.

But the fear still left marks, and the marks have a body count. Golden Rice was tied up for years while children went blind from exactly the deficiency it was engineered to prevent. The pause on gain-of-function research may have left the country less ready for COVID, not more. Europe regulated its agricultural biotech into the ground until the companies and the scientists simply moved to places that would have them. Not pursuing a technology is not a neutral act. It’s choosing the status quo over a possible improvement, and the status quo has victims too. They’re just diffuse, and unphotogenic, and they never testify.

I want to be careful here, because the easy misreading is that risk is fake. It isn’t. The forty-thousand-molecule experiment was a genuine wake-up. The point is narrower and harder to hold: the debate gets skewed by the most vivid worst case in the room, and the correction is to right-size the risk against evidence instead of against imagination.

And when you go looking at who’s holding the vivid worst case, the fear usually turns out to have beneficiaries. Not a conspiracy — incentives. A doomsday framing sets research agendas, wins funding, concentrates attention on whoever raised the alarm. That doesn’t make the alarm insincere. It just means that when someone shows you a monster, it’s worth asking who’s doing the showing, and what they’d like you to do while you’re frightened. That question sits under the testimony, under the essay, and, if I’m honest, under the novels too.

There’s a line I wrote into the last book of Echoes of Tomorrow. A character stands over a ruined Earth and explains what actually killed it — not a weapon, not a plague, but “bureaucrats who refused to take bold action because they were afraid of making a mistake.” Fear killed Earth. I put that in a novel, and then I went and said the sober, un-apocalyptic version of it to four senators. Same fear, two registers.

The essay ends on a phrase I meant as a personal doctrine, and I’ll stand on it here. Cautious, but unafraid. That’s the whole posture. You don’t get certainty in either direction — not the utopian kind that says build it all and don’t look back, not the apocalyptic kind that says freeze until we can prove a negative. What you get is a baseline to measure against, and the nerve to keep moving while you read it.

The forum was called Guarding Against Doomsday Scenarios, and I still think that’s exactly the right assignment. I just think the list they walked in with was one scenario short.