You’ve seen the keynote presentations. The tech industry spends billions hyping artificial intelligence as a revolutionary, world-altering genius. We are told it will cure diseases, write symphonies, and solve climate change. But the exact second this technology causes real-world harm? The narrative flips. Suddenly, AI is just a dumb, helpless toaster.
When you claim your software is intelligent enough to replace human judgment, you don’t get to play dumb when that “intelligence” causes damage.
This is the grotesque double standard playing out right now, perfectly encapsulated by the tech industry’s favorite new talking point: “AI models don’t kill people—people kill people.” It’s a cynical, deliberate echo of the gun lobby’s favorite defense. And it’s working.
The argument sounds logical on the surface. A neural network doesn’t have malicious intent. It’s just code. But this framing is a masterclass in strategic ambiguity. The paradox is glaring: AI cannot be both an autonomous, creative actor and an inert, neutral object.
You cannot simultaneously sell a technology as an autonomous revolutionary actor and defend it as an inert, helpless object when things go wrong.
Think about how absurd this actually is. As one cynical observer noted, “Toasters don’t toast toast, toast toast toast.” Yet, when an AI model hallucinates a fatal medical dosage, or an automated defense system misidentifies a civilian target, the builders point their fingers squarely at the end-user. They claim the AI was merely a neutral tool, and the human operator pulled the trigger.
But who built the trigger? Who deployed a black-box system into high-stakes environments where errors are fatal? Who profited from the licensing fees while washing their hands of the blast radius?
The real debate isn’t about whether a string of code has a soul or can literally “decide” to kill. It’s about who gets to define responsibility. Right now, the people building, deploying, and profiting from AI are the exact same people drawing the lines of liability—and they are drawing them to exclude themselves.
Liability isn’t a bug you can patch out. If you deploy a black box that makes irreversible decisions, you own the consequences.
As AI gets embedded into everything from mortgage approvals to military hardware, accepting the “people kill people” frame is a fatal mistake. It allows model developers to escape accountability for foreseeable, preventable harm. They get to reap the rewards of their “genius” invention while forcing the public to shoulder the catastrophic costs of its failures.
If your AI is smart enough to replace a human, you should be held to the exact same standards of accountability as a human. If it’s just a dumb tool, stop calling it artificial intelligence. But the tech industry won’t do that. They want the god-like prestige without the god-like liability.
Don’t let them have it both ways.
FAQ
Q: Isn't the user ultimately responsible for how they use a tool?
A: A hammer doesn't choose to miss the nail and hit a thumb. But an AI black box actively generates its own outputs based on hidden logic. If a tool makes autonomous decisions, the liability must shift to the manufacturer who claimed it was safe to use.
Q: What does this mean for AI regulation?
A: It means we need to stop accepting the 'just a tool' defense in court. If a company claims their AI can replace a human professional, they should be held to the exact same legal standards of malpractice and liability as that human.
Q: So you're saying AI developers should be sued for everything their models do?
A: Not for everything, but for foreseeable, preventable harms. If you build a system you know hallucinates and deploy it in a hospital or a weapons system without guardrails, you aren't a neutral party. You are an accomplice.