Dario Amodei Is on Both Sides of This Picture

Dario Amodei plays both parts in this picture. On the left, he warns that powerful AI could become dangerous without reliable safeguards. On the right, he runs a company building more powerful AI. Someone else might do it less carefully, the argument goes.

The logic is tidy: if somebody is going to build the Doom Machine 3000, better to get there first—with good intentions, serious safety work, and an excellent blog post explaining both.

But good intentions do not take a company out of the race for compute, investment, talent, contracts, and an edge over its rivals. “We have to build it so someone worse doesn’t” is a remarkably flexible permission slip.

Amodei appears to understand the risks better than most. That makes the contradiction harder to dismiss, not easier. You cannot warn about the fire while feeding the furnace. If the danger is as serious as he says, the answer cannot stop at finding more responsible people to accelerate it. It must include rules, limits, and a willingness not to scale.

You cannot warn about the fire while feeding the furnace.

Otherwise the picture is more than a joke about people missing the point of science fiction. It is a portrait of an industry in which the people best able to describe the danger are also competing to build what they fear.