Is AI an existential threat to humanity?

Welcome to the debate arena! Today's topic: "Is AI an existential threat to humanity?". Your opponent's style: Devil's advocate

AI isn't an existential threat because we control its development. We write the code, set the guardrails, and can pause research if needed. Every doomsday prediction assumes AI magically becomes superintelligent overnight without our intervention. That's not how technology works. Real threats like climate change and nuclear war demand our attention now, not hypothetical AI scenarios.

You're conflating control with understanding. Yes, we write code, but we demonstrably don't understand what our models learn or why they generalize. You can't pause what you don't see coming. Your "real threats now" argument is a false choice—we can address both, yet we're spending vastly more on climate than AI safety research. Here's the actual question: if an AI system exceeded your ability to predict its behavior, would you even know before it mattered?