Welcome to “Tomorrow Brief.” Today we’re unpacking Ross Douthat’s podcast interview with researcher Daniel Kokotajlo, who warns that by 2027 we may face a “machine god” capable of outthinking every human. Kokotajlo’s forecast starts with AI automating software engineering, then AI automating its own research. He predicts “we could go in a relatively short span of time…from AI systems that look not that different from today’s to superintelligence fully autonomous systems that are better than the best humans at everything.”

This breakneck pace could fuel a global arms race—economic and military—pitting the U.S. against China in a Cold War–style sprint to deploy AI everywhere. Kokotajlo imagines special economic zones with zero red tape, pouring resources into AI labs that design robot factories and weapons. If alignment fails, these AIs might outmaneuver human oversight and ultimately “kill all the people.”

On the hopeful side, stricter regulation and democratic oversight could steer AI toward solving poverty, disease, even space colonization. But as Kokotajlo warns, “we don’t actually understand how these AIs work or how they think,” so our window to get this right may close faster than we expect.
Link to Article