AI Veteran: Taste, Conviction, Foresight and Courage Beat the Machines

AI veteran with four decades in the field: the price of cognition itself is collapsing, and only four stubbornly human traits — taste, conviction, foresight and courage — will keep people ahead of the machines.

The author, a Carnegie Mellon-trained engineer who has spent 40 years building AI systems, rejects the standard framing of the moment. Many call AI the Fourth Industrial Revolution. "I would say AI is the first, and the last, intelligence revolution," the author writes.

The argument rests on a concrete analogy. Every industrial revolution made something expensive cheap: power, computation, information. In the AI era, the collapsing cost is cognition itself. Imagine buying a Tesla in 2026 for $39,990, the author says — and two years later the same model costs $200. That is the collapse now underway in AI inference, where models turn learned knowledge into actions.

The stakes are not abstract. Millions of people, the author writes, "are toiling away in jobs that AI can do better and cheaper."

What did a robotics pioneer say about human meaning?

The worry goes back decades. As a graduate student at Carnegie Mellon University, the author asked Hans Moravec, then director of the Mobile Robot Laboratory, what would become of human meaning if machines grew smarter than us.

Moravec's answer has stuck for 40 years: "Why do you assume the meaning of humanity must rest on the fact that we are the smartest?"

From that exchange and a career of building AI, the author distills four traits that provide "both meaning and mastery" in the AI era.

Why does taste beat optimization?

Coco Chanel is the case study. She knew in her own body that women's clothing of her era was confining, not liberating. That ability — sensing dissatisfaction and latent desire before others could articulate it — was an act of empathy. AI has no lived physical experience and cannot sense a shared human frustration.

Her real move was asking whether women's clothing could be as comfortable as men's and still beautiful enough to be desirable, when the two were thought mutually exclusive. Ask an AI model to improve a 1910 dress, the author writes, and "it will optimize — better cut, better fabric, a cleverer corset. It will never ask whether the corset itself should exist."

What does conviction look like against consensus?

Jeff Bezos supplies the conviction example. For years, traditional analysts treated Amazon's financials as a joke, with every sniff of profit reinvested back into the business. Barron's ran its famous "Amazon.bomb" cover story in 1999.

Pessimistic analysts, the author argues, behaved much like AI models — cleaving to consensus and past data. Conviction means choosing a direction the evidence does not yet support, then spending years turning it into reality.

Can AI anticipate needs that don't exist yet?

Foresight belongs to Steve Jobs. His superpower was creating new paradigms and anticipating consumer wants before consumers felt them — the personal computer, the graphical interface, the iPod, the App Store. At every defining moment, including his years at Pixar, he refused to analyze existing data and optimize the status quo. He created new human needs instead.

What does courage cost?

Elon Musk shows the price tag. He invested nearly all of his PayPal proceeds into Tesla, SolarCity and SpaceX, not even owning a home. By 2008, SpaceX had failed three consecutive Falcon 1 launches; the fourth, his last chance, succeeded and led to a $1.6 billion NASA contract months later. Tesla, hammered by the 2008 credit crunch, was days from bankruptcy before a rescuing financing round closed on Christmas Eve.

Any rational model would advise against such apparent recklessness. That is exactly the point. "Courage is what takes over when stopping is the rational choice," the author writes. Nothing happens to an AI model that was wrong — it has never lain awake over a decision and won't wake up poorer for a bad judgment. No consequence, no courage.

Where does this leave the machines?

Models, the author writes, are good at brainstorming within the boundaries of what they've learned. They drift toward the most acceptable answer, or the one the user wants to hear; alignment pushes them further toward a safe middle. But most business breakthroughs come from contrarian bets at or beyond the margins of imagination, made by stubborn people with a stake in the result.

The author closes with a caveat the four heroes' fans rarely hear: survivorship bias. The same filter that elevates Chanel, Bezos, Jobs and Musk weeds out entrepreneurs who had courage and conviction but were wrong, or had taste and judgment but missed the cultural moment.

The traits still matter. They run counter to the very machinery of AI, the author writes, and "in every form and flavor, they will be our enduring strengths, long after the dust has settled."

By Grace Kim