Tech Veteran Pushes Back on AI Doomsday Warnings, Calls for Human Oversight
A 20-year industry veteran argues that artificial intelligence is a tool like the internet or the smartphone, not an existential threat, and that the real challenge is how humans train and govern it.
Artificial intelligence is not an existential threat but a tool whose risks stem from how people use it, a technology industry veteran with more than two decades of experience argues, pushing back against warnings from major tech companies about AI-driven catastrophe. The core problem, the argument goes, is not the technology itself but the humans who train, deploy, and sometimes misuse it.
In recent years, prominent technology firms have warned that AI could take control away from society, eliminate jobs, enable bad actors to design dangerous pathogens, trigger a catastrophic cyberattack, or destabilize financial systems. The veteran dismisses those scenarios as fearmongering, comparing AI to earlier technological leaps — the internet, pagers, cellphones, and smartwatches — that were initially met with suspicion but ultimately delivered broad benefits.
The comparison to the internet is central to the argument. The internet gave the world nearly unrestricted access to information, transformed how people work, and created opportunities for engagement that did not previously exist. It also gave criminals, terrorists, and other bad actors new ways to collaborate and evade systems. Yet most people would still describe the internet's rise as an overwhelmingly positive development for society. AI, the argument continues, is following a similar trajectory.
The veteran stresses that humans remain involved at every stage of AI development, even in cases where models have erroneously hacked government websites or escaped their testing environments. Because AI lacks a moral compass and cannot weigh ethical dilemmas, the people training these systems must apply strong guardrails — described as constitutions — to ensure the technology complies with the moral standards society expects.
Large language models, the systems that have become household names worldwide, are trained to produce probabilistic responses. In practice, that means they generate the answer they calculate the user most likely wants, which frequently leads to hallucinations — industry jargon for fabricated or false information. Human users and governance structures provide a necessary check, which is why people must continue to verify facts and check sources rather than treating AI output as authoritative.
The veteran argues that the public debate has focused on the wrong question. Instead of asking whether AI is good or evil, beneficial or harmful, society should ask who is training AI models and for what purposes, what capabilities the technology may develop in the coming decade, and how it will reshape industries. Healthcare, finance, and manufacturing are among the sectors expected to see significant impact.
The argument also rejects the idea that AI is designed to take jobs or control people's lives. It is described instead as another tool in the toolkit — something to reach for when an extra set of hands is needed, when work needs to be sped up, or when tasks can be simplified. The veteran recalls an elementary school math teacher who insisted the subject mattered because students would not carry calculators to the grocery store, the gas station, or the department store. A couple of decades later, nearly everyone carries a calculator on their phone.
The conclusion is a call to set aside fearmongering and recognize the benefits AI can bring, while continuing to develop models responsibly with human involvement at the center. The hope, the veteran writes, is that the public will fully grasp those benefits and close the door on alarmism once and for all.
8
