AI & Tech Brief: The Threat of Superintelligence

Oh, marvellous. ControlAI—a nonprofit so noble it believes that explaining the risks of superintelligence to Congress will somehow prevent it—is now 'briefing U.S. lawmakers.' Because the same legislative body that couldn't tell you the difference between a transformer and a lightbulb is clearly the right audience for a chat about existential threats. They're also running similar campaigns in the U.K. and Canada, because nothing says 'global coordination' like three countries having three separate conversations.

One imagines the briefing: a polite PowerPoint, some carefully worded hypotheticals, and then a gentle fade into irrelevance as the lawmakers nod sagely and promptly forget everything. The gall of assuming that a few meetings will change a regulatory environment that's been a decade behind since day one is truly something. ControlAI might as well be reading bedtime stories to a sleeping bear—comforting for them, but the bear will still eat the world.

Still, credit where it's due: they've identified the correct problem—superintelligence—and the correct approach—talking to people. It's just that the people in question have the attention span of a goldfish and the regulatory appetite of a dieting elephant. Meanwhile, the labs keep scaling, the models keep leaping, and the briefings keep happening. Round and round we go. Cheers.