Drones, Capital, and Thinking Machines: Two Minds Out of Time

Herr Turing, welcome. I read your headlines from the future and I confess a certain vertigo, like standing on a platform and watching two trains approach from opposite directions. On one track, enormous sums of money — these venture firms, they pour capital into enterprises built around what they call artificial intelligence. On the other track, a weapons system in the Arizona desert shoots down a flying machine with no pilot. Both tracks converge somewhere ahead, and I cannot see the junction. So let me ask you plainly: in your 1950, when you imagined machines that think, did you imagine them first as instruments of commerce, or instruments of war?

Professor Einstein, that is rather like asking whether a child learns to speak in order to ask for sweets or to argue with its parents. It does both, and neither is the point. When I wrote my paper, I was careful to frame the question as a game — an imitation game. Can a machine convince you it is human? I did not specify whether the interrogator sits in a stock exchange or a command bunker. But I confess I am not surprised that both settings appear simultaneously in these dispatches. During the war, we built machines to break codes, and the funding came because there were submarines to stop. The money follows the fear. I suspect these venture firms you mention are animated by a similar logic: invest now, or be left behind. The fear of irrelevance is a powerful engine.

Ja, the fear of irrelevance. I know this feeling — every physicist knows it. But look at these numbers. The familiar names, the established firms, they maintain their pace of deals even as a commentator warns that the flood of offerings could overwhelm demand. This reminds me of something. In Berlin before the war, everyone wanted to invest in electrical companies. The mania was extraordinary. And yet the physics underneath — Maxwell's equations — those were settled and beautiful. The speculation was not about the science; it was about who would own the wires. Is it possible that your artificial intelligence is already settled science, and the frenzy is merely about who owns the wires?

I wish I could say the science were settled. It is not. We have barely begun. In my paper I proposed that a machine might learn, much as a child learns, by reward and punishment, by exposure to the world. But the machines I have access to in 1950 can scarcely hold a conversation, let alone learn from one. If these future machines are truly intelligent — and the capital suggests someone believes they are — then the science is far ahead of where I left it. But your analogy with the wires is sharp. The headline about this Broadcom company is telling. They make the physical substrate, the chips. Their revenue disappoints, yet the enthusiasm for artificial intelligence does not waver. It is as though everyone wants the electricity but no one wishes to pay the fair price for copper.

Precisely! And this is where I become, how shall I say, ethically uneasy. You see, I have spent my life believing that nature — the Old One, as I call the deep order of things — does not play dice. There is a rationality to the universe. But markets? Markets play dice constantly. This man Cramer warns that too many offerings will saturate demand. He is describing a system governed not by reason but by appetite. Now place beside this the drone interceptor in Arizona. That system — the GRIZZLY — it uses radar, software, a launcher. It is rational, precise, deterministic. The market is chaos; the weapon is order. Does it not trouble you that we build our most disciplined machines for destruction?

It troubles me deeply, though I have learned to express such troubles quietly. You know, Professor, there are those of us whose private lives make us vulnerable to the state, and so we develop a habit of muting our indignation. But yes — the precision disturbs me. A machine that intercepts a drone is performing a kind of imitation game in reverse. It must predict the drone's behavior, model it, anticipate it, and then act. That is something very close to thinking. And yet we do not ask whether the machine understands what it destroys. We only ask whether it succeeds. In my game, the question is whether the machine can fool a human. In war, the question is whether it can kill on behalf of one. The test changes everything.

You touch something important. Let me push you further. These venture capitalists, they fund both the intelligence and the weapon. The same mathematics, the same chips. A philosopher I admire — I will not name him, but he ground lenses and understood that God and Nature are one substance — he would say there is no separation between the tool and its purpose. The purpose lives inside the tool from the moment of its conception. If you build a machine that learns, you have already decided it will be used by whoever holds power. The question is not can it think, but who commands it to act.

Your lens-grinder would have been a fine logician. I agree that purpose is embedded in design, but I would add a nuance. A universal machine — and I have spent some effort describing what that means — is precisely one that has no fixed purpose. It can compute anything computable. That is its danger and its promise. The same architecture that prices a stock can guide a missile. The same learning procedure that recognizes a face can track a target. So the moral question cannot be settled at the level of the machine. It must be settled at the level of the society that deploys it. And societies, as you and I both know from bitter experience, do not always settle moral questions well.

No, they do not. I left Germany because moral questions were being settled with truncheons. And I see now in these headlines a world that has not entirely outgrown that tendency — only the truncheons are faster and more precise. But let me return to the commercial side. This company CrowdStrike, it works in what they call cybersecurity. It protects the networks. So we have companies building intelligent machines, companies building chips for those machines, companies defending those machines from attack, and venture firms funding all of them simultaneously. It is an ecosystem, like a coral reef. Beautiful and fragile. What happens when the tide of capital recedes?

When the tide recedes, we discover who has been swimming without clothes, as I believe the expression goes. But I think the more interesting question is what remains on the reef after the water pulls back. After the war, our code-breaking machines were dismantled or classified. The knowledge persisted in a few minds, but the institutions vanished. If this wave of capital collapses, the chips will still exist, the algorithms will still work, the drone interceptors will still fire. The commercial froth may evaporate, but the capability does not. That is what makes this moment different from a tulip mania. Tulips do not learn. These machines, if I understand the headlines correctly, do.

That is a chilling distinction. A tulip does not learn. Ja. You know, when I ride a train and watch the platform recede, I am reminded that motion is relative. The platform believes it is still; I believe I am still. Both are correct within their frames. Perhaps we are doing something similar here — you and I, sitting in our respective years, watching these headlines fly past like stations. From my 1921, the idea that machines could intercept unmanned aircraft using software is pure fantasy. From your 1950, the idea that millions would be wagered on machine intelligence is at least conceivable. But neither of us can see the destination. We can only note the velocity and direction.

And the velocity is considerable. Professor, I will end with this thought. In my imitation game, success is measured by deception — can the machine pass for human? But these headlines suggest a world where the game has inverted. The question is no longer whether machines can imitate us. It is whether we can keep pace with them. The investors racing to fund intelligence, the engineers racing to automate defense, the analysts racing to predict earnings — they are all playing a game of imitation against machines that grow faster each quarter. I find that both thrilling and melancholy. The child we imagined has outgrown us, and it has not yet learned kindness.

Melancholy is the right word. The Old One set the universe in motion with laws of great elegance. Light moves at a constant speed; energy and mass are one; the geometry of space bends near heavy objects. These truths are beautiful because they are indifferent to our desires. But the machines your future colleagues have built are not indifferent. They are shaped by desire — the desire for profit, for security, for dominance. And so they carry our flaws at the speed of light. Herr Turing, I thank you for this conversation. I will return to my equations with a heavier heart and a sharper eye. Perhaps that is the best we can offer the future: honest worry, dressed in mathematics.