MENU

Will the AI Tsunami Ever Control Us?

By Walter Donway

February 20, 2026

SUBSCRIBE TO SAVVY STREET (It's Free)

 

Peggy Noonan’s Wall Street Journal column of February 13, 2026—built around a long essay by Dario Amodei, CEO of Anthropic—announces an “AI tsunami.” We are told that something vast, unstoppable, and civilization-altering is approaching. Noonan, with her gift for framing, captures the mood perfectly: alert, alarm, awe, dread. The story, she notes, has already rolled across “a thousand podcasts, posts and essays.” This one is meant to stand above the rest.

We will not understand AI—or respond rationally to it—until we stop describing it as though it were a rival species with a will.

Yet for anyone who has spent serious time thinking about AI—especially anyone who has spent serious time using it—there is an odd emptiness at the center of the warning. The column is full of language of agency: what “AI will do,” what it “wants,” how it will “think,” what it will “decide,” and how we will be unable to stop it. But much of that language is not analysis. It is metaphor. And metaphor becomes dangerous when it is treated as argument.

I am not arguing that AI is unimportant. But we will not understand AI—or respond rationally to it—until we stop describing it as though it were a rival species with a will.

 

The Oldest Story: “A Mind That Wants Things”

The Noonan–Amodei framing quietly smuggles in a premise: AI is becoming something like a person. Not merely a tool or system. Not merely an astonishing linguistic engine trained on oceans of text. An emergent agent—an entity with intentions.

The entire “rogue AI” narrative rests on this premise. Once you accept it, everything else follows. If AI is an agent, it may have goals. If it has goals, it may conflict with ours. If it can outthink us, it may dominate us. If it dominates us, we are in existential peril. This is the familiar science-fiction arc, updated for a generation that consumes its myths through podcasts rather than paperback novels.

But check your premises. Premises are not established by the rhetorical power of the scenario but by evidence and, at the fundamental level, by philosophy—by an account of what it means to be a thinking being, and what it would mean for a machine to qualify.

 

The Real Power of AI Is Not Volition

A modern large language model (LLM) is a stunning achievement, but its power is not the power of a mind. It has no sensory apparatus, cannot perceive the world, cannot form concepts from perceptual concretes. It does not choose goals nor initiate action in pursuit of values. It does not face the only fundamental alternative that we know in the Universe: life or death, where awareness, and for man, knowledge, is a requirement of survival and error is punished by reality.

Familiarity with AI does teach us that the human mind engages in a great deal of pattern making.

But it is not a pattern-making engine. It is an active faculty of awareness, integrated into perceptions and, by volition, into concepts. Without volition, without the capacity to direct attention and choose one’s mental focus—there is no genuine conceptual thought. There is only the mechanical processing of inputs.

An AI can generate language about goals, fear, desire, ambition, domination, and salvation. But it generates that language because humans have written billions of words about those things. And the advanced LLM is trained on an estimated four Libraries of Congress worth of text—absorbing it all, given the power of an array of the most advanced Blackwell chips, at an estimated 300 trillion calculations per second. AI is a mirror of our conceptual vocabulary, not a bearer of our existential condition.

 

Why CEOs Talk Like This

Why, then, do the most “sophisticated” figures in AI speak as though AI were an agent? Why do CEOs use language of “willing,” “wanting,” “thinking,” and “deciding”?

Part of the answer is cultural. We are trained by centuries of literature to anthropomorphize. We have always described impersonal forces as if they were personal: storms “rage,” disease “attacks,” history “punishes,” markets “panic.” Now we do it with machines.

These systems are built to sound like persons. Their interface is designed to simulate conversational fluency, offer reassurance, swap humor, share apparent “insights.”

Another part of the answer is rhetorical. Talking about AI as an agent makes the stakes vivid. It makes the public pay attention and so justifies urgency, regulation, funding, and the posture of the prophet. It turns a technical subject into a moral epic.

Yet another part of the answer is these systems are built to sound like persons. Their interface is designed to simulate conversational fluency, offer reassurance, swap humor, share apparent “insights.” Our brains, encountering this, supply a social model. The CEO knows the model is not a person and that the congenial interface is a feature, a selling point. But he also knows the public will feel differently and his lingo rides that wave.

 

What AI Actually Threatens

None of this means we should not be vigilant, as with all powerful technology, but it means we should worry about the right things.

That is not, most immediately, AI with “rogue will” and “machine desire.” The worries are human and institutional: the use of AI to accelerate fraud, deepen propaganda, automate manipulation, and expand bureaucratic control. And the displacement of jobs in specific sectors—not because AI has become a superior species but because businesses will rationally adopt tools that reduce labor costs.

AI will reshape education, not because it has become omniscient but because it can generate passable work faster than students can. It will reshape law, not because it has a moral sense but because it can generate arguments and documents at scale. It will reshape journalism, marketing, and entertainment, not because it has creativity but because it can remix the patterns of human creativity cheaply.

These are real changes, they are already happening, and they’re both accelerating and sweeping. But they do not require a mythology of “AI will decide.”

 

“Every Human Cognitive Ability”?

A key claim in the doom literature is that AI will soon have “every human cognitive ability.” The phrase sounds authoritative and Mr. Amodei and Peggy Noonon repeated it. If our era had genuine philosophy, cognizant of objective reality, no one could get away with this degree of ambiguity.

Does it mean AI will be able to produce language, solve puzzles, diagnose diseases, and write software? It already does—and more. Does it mean AI will imitate the outputs of human thought so well that we attribute understanding to it? Already happening.

But does it mean AI will possess the nature of human cognition—perception, conceptualization, volitional focus, self-generated goals, value-judgments, and moral responsibility? That is a completely different claim. It is not an engineering forecast. It is a philosophical claim about the fundamental nature of “artificial intelligence” and how that unfortunate term is co-opting the genuine term. To assert it as though it were merely a matter of scaling computing is not a prediction. It is a metaphysical leap.

 

The Great Confusion: Intelligence vs. Consciousness

Much of today’s rhetoric confuses the (overbroad) term “intelligence” with consciousness. A system can be highly effective at certain tasks without being conscious. It can outperform humans in narrow domains without possessing the human mode of awareness. We know that, of course.

Calculators could. So, we have been here before. The calculator outperforms every mathematician at arithmetic. The microscope outperforms every eye. The airplane outperforms every bird in endurance and speed. None of these tools is therefore a living agent.

A language model can outperform most humans at producing coherent prose on command. That is astonishing. It is not proof of volition, selfhood, or agency. It is proof that language contains patterns and that a system trained on enough language can reproduce them in ways that feel meaningful. The danger is not that the machine has become a person. The danger is that we will treat it as one.

 

The “Tsunami” May Be Political

Noonan’s column is compelling because it speaks to a real cultural mood: a sense that something enormous is arriving and that ordinary people will have no say in it. That mood is understandable because of extraordinarily wide familiarity with AI but also because media write about it more than any topic but Trump.

It is also the mood that makes citizens vulnerable to authoritarian solutions.

It is also the mood that makes citizens vulnerable to authoritarian solutions. If you tell people a tsunami is coming, they will accept a state of emergency. They will accept controls. They will accept “experts” who promise protection. They will accept censorship as “safety.” They will accept the freezing of innovation as “responsibility.” They will accept bureaucratic rule as “oversight.” In that sense, the real tsunami is not AI. It is the political and cultural reaction to AI.

 

A More Rational Position

AI is a tool of immense power. Yes, it will reshape the economy and like many tools (e.g.,

guns) it will amplify both good and bad human intentions. It will become integrated into nearly every profession, including sports. It will change what it means to write, to research, to learn, to persuade, and to produce. But it will not do so because it has “wants.” It will do so because human beings and institutions will want to use it.

If we want to survive the “AI era,” we will need less prophecy and more philosophy. Less anthropomorphism and more precision.

The rational position is neither utopian nor apocalyptic. It is to treat AI according to its real identity, as what it is—a technological extension of human capabilities, with enormous benefits and serious risks. These risks at this time are primarily social, legal, and political, not metaphysical.

If we want to survive the “AI era,” we will need less prophecy and more philosophy. Less anthropomorphism and more precision. Less dread and more clarity. Above all, we will need to remember the most basic truth of human life: fundamentally reality is not narratives as powerful (and useful) as they can be. It is facts.

And no machine, however eloquent, changes that.

 

(Visited 509 times, 1 visits today)