| He is not as famous as Dario, Elon, Sam and the other AI leaders known by their first names. But Eric Schmidt, the former CEO of Google, is playing an outsize role in miltech, the industry that will shape the future of war. He visited Ukraine, including the front line, early in the conflict and decided to help—building and scaling a drone company whose products have been deployed at, and behind, the front lines. I first met Mr Schmidt more than a decade ago (when he was a director of the The Economist Group) and recently had the chance to catch up with him in Kyiv. Tomorrow’s Inside Defence will be the final instalment in a trilogy of episodes filmed in—or about—Ukraine, following our panel discussion unpacking the lessons from our visit and a special report from inside an elite Ukrainian drone unit. This episode goes beyond the battlefield, though, and grapples with broader questions around this new, dystopian era of AI warfare.
Our conversation started with the situation in Ukraine. Mr Schmidt gave a strikingly sombre assessment. He warned that Russia’s pace of innovation is accelerating and he made a powerful case for why Ukraine needed financial support and help to scale its own innovations faster. Intriguingly, he also argued that the American government could order SpaceX to allow Ukraine to use Starlink over Russian territory in order to hit the launchers of the ballistic missiles that are terrorising Ukrainian cities and that, generally speaking, he’d support such a move.
We moved onto a wider discussion about the impact of ever more powerful AI on warfare—especially the transformative improvements in planning, efficiency and execution that could lead, as Mr Schmidt put it, to a grim hypothetical scenario in which a dictator tells its “perfect” AI system: “Attack this country. Tell me when you’re done.” We were speaking a few days after a researcher from Anthropic, a leading AI company (and one in which Mr Schmidt was an early investor) warned that AI “could kill us all by the end of the decade”.
Mr Schmidt stressed the importance of establishing some “basic rules about AI in war”. But how? I’ve become increasingly pessimistic about whether this is possible or, indeed, whether it will be possible to ensure the safe development of the technology amid an almighty AI arms race between America and China. The war in Ukraine shows how rapidly the innovation cycle advances in the AI era, how much even a small technological edge matters in wartime—and what happens when, in the middle of an existential battle, one side doesn’t play by the rules. The result is a relentless cycle of innovation that threatens to spiral out of control. If “whoever wins AI, wins”, as President Donald Trump puts it, it will surely be a struggle for America and China to agree on norms that constrain their pace of innovation.
These questions were at the heart of last week’s cover package and we discussed them at length. Mr Schmidt’s position was a striking combination of fatalism and optimism. He did not think there would be any agreement to slow down the deployment of AI (“there’s not going to be a pause,” he said, because it was unverifiable and “counter to the incentives of all players”). But there would be alternative ways to ensure safety (“You have to give the other side a chance to say, ‘You’ve gone too far’”). And he was sure that norms would be developed for the use of AI in warfare. “As long as AI is under human control and under human alignment,” Mr Schmidt said, we will be OK.
I hope he is right.
I’m keen to hear what you think. It was a conversation that sharpened my thinking on some of the biggest and most complex issues we now face, and I hope it does for you, too. Once you’ve watched tomorrow’s episode, please write to me at insider@economist.com with your thoughts and feedback. The interview will be available to watch from 6pm London time (1pm in New York) on Tuesday. |
