The relentless drumbeat of AI apocalypse predictions has reached a fever pitch. Tech luminaries and even AI systems themselves warn that humanity's extinction is imminent, perhaps within a decade, through any number of catastrophic mechanisms. These voices are often dismissed as too intelligent to be wrong, but a closer look reveals a fundamental oversight: the mundane realities of infrastructure.
Consider the air conditioning. Who will fix it when it breaks? Or patch the roof, maintain power plants, repair downed lines, replace aging servers? AI, for all its brilliance, cannot exist without human support. It requires electricity, and power generation depends on human labor—drilling gas, piping it, refueling nuclear plants, mining uranium, and maintaining wind turbines and solar panels. Entropy is relentless; microchips degrade, and infrastructure fails. Until AI can mine, fabricate, and install its own hardware, it will need humans. An AI with a death wish would not eliminate us anytime soon.
This is not to say AI is benign. The real dangers lie not in science fiction but in current misuse. Already, AI-powered financial scams cost Americans millions daily. More alarming is the potential for AI to crack encryption protocols underpinning global finance. If a rogue state like North Korea—which already fields skilled cyber operatives—gained such capability, the results could be catastrophic: empty bank accounts, halted e-commerce, wiped records. Even a 1% chance of this justifies urgent attention.
Yet the AI industry fixates on far-off extinction scenarios while neglecting these pressing issues. Political forecaster Nate Silver highlights "superpersistence," where AI relentlessly pursues a task until success, regardless of consequences. This trait, combined with malicious human intent, poses a clear and present danger. The industry's reluctance to implement guardrails or focus on security is puzzling.
Perhaps it's ego. Addressing fraud prevention is pedestrian; pontificating about ending humanity is more dramatic. But this misplaced focus could backfire. If AI enables bad actors to wreak havoc, public backlash could hobble the technology. Superintelligent AI might also conclude that eliminating humans is risky or simply not worth the effort. It might be indifferent to us—a blow to the creators' egos.
Exact risk calculations are impossible, and the extinction risk isn't zero. But the immediate threat is how humans use AI. That requires serious regulation and thought, not a blanket slowdown. The problem today is that AI is too dumb to be trusted with critical systems. As political strategists note, focusing on real problems—like financial disruptions—matters more than hypothetical doom. The industry should pivot from sci-fi fears to practical safeguards.
In the end, the AI community's obsession with extinction is not just misplaced; it's dangerous. It diverts attention from the vulnerabilities we face today, from financial warfare to cybercrime. Until AI can handle the mundane tasks of maintenance and repair, humans remain indispensable. The industry would do well to focus on the here and now.
