There is an old rule in politics and business: follow the money.
It is worth remembering as America enters its increasingly heated debate over artificial intelligence. AI is not simply another technological innovation. It is likely to reshape work, education, medicine, warfare, communication, finance, and perhaps our understanding of what it means to be human. Some apprehension is therefore entirely reasonable.
Fear, however, is an unreliable guide.
Fear can sharpen our attention to genuine danger. It can also distort our perception of it. And in a society with declining institutional trust and little shared cultural consensus, measured responses are increasingly difficult. We retreat to our corners. Complexity becomes conspiracy. Risk becomes apocalypse.
That is beginning to happen with AI.
There are legitimate reasons to be concerned about increasingly autonomous AI systems. There are legitimate questions about cybersecurity, fraud, surveillance, employment, military applications, misinformation, privacy, and the possibility that highly capable systems may behave in ways their designers did not anticipate.
These questions deserve serious attention.
But serious attention is precisely what makes hyperbole difficult.
The public debate increasingly presents us with a false choice: either embrace AI without reservation or fear that autonomous machines are about to bring civilization to an end. Neither posture is adequate at the moment.
The more interesting question is why the alarm has become so intense now.
Part of the answer may be found by following the money.
The stakes in the AI industry are almost unimaginably large. Anthropic is reportedly preparing for an initial public offering that could value the company at roughly $2 trillion. Other frontier AI companies are attracting capital at levels that would have seemed fantastical only a few years ago. Billions are being spent on chips, data centers, electricity, talent, and model development.
When that much money is at stake, debates about safety cannot be neatly separated from debates about competition, liability, regulation, and market position.
The emerging AI giants have legitimate reasons to want clear rules. But regulation also has economic consequences. Rules that a trillion-dollar company can absorb may become insurmountable barriers for a smaller competitor. Regulation can protect the public. It can also protect incumbents.
That distinction matters.
Liability matters as well. The more potentially consequential the technology, the greater the incentive for companies to establish clear governmental standards governing its use. Such standards may be necessary. But they can also distribute responsibility. If something eventually goes badly wrong, companies operating within a government-approved framework are in a very different legal and political position from companies left entirely responsible for policing themselves.
None of this means AI executives are insincere when they warn about risks. It means that moral, technological, financial, and institutional incentives are intertwined. We should be sophisticated enough to recognize all of them at once.
The same complexity applies to data centers.
Data centers have become the physical symbol of the AI revolution. They are, in effect, the industrial libraries of the digital age: enormous concentrations of computing power that store, process, and increasingly generate information.
They require land. They require infrastructure. Above all, they require electricity.
And some require substantial amounts of water.
These are real issues. Communities considering massive data-center developments have every right to ask who pays for new transmission lines, where the electricity will come from, how much water will be consumed, what happens to utility rates, and whether promised economic benefits justify the costs.
But again, the details matter.
Not all data centers consume water in the same way. Some rely heavily on evaporative cooling. Others increasingly use air cooling, reclaimed water, or closed-loop systems that dramatically reduce freshwater consumption. Amazon, for example, reports using air cooling for most of the year and water primarily during the hottest periods. Other companies are developing closed-loop systems designed to avoid continuous consumption of potable water.
The statement “data centers use too much water” tells us surprisingly little. Which data center? Where? Using what cooling system? Drawing from what water source?
Electricity is generally the larger structural challenge.
AI requires extraordinary computing power, and extraordinary computing power requires extraordinary amounts of energy. That will require changes in America’s energy infrastructure. But technological civilization has always required new infrastructure. The automobile eventually required interstate highways. Commercial aviation required airports. Electrification required a national grid. The Internet required fiber-optic networks. This power need also creates the source of new innovations that can reduce power needs of data centers by as much as a sixth.
AI will require its own physical infrastructure.
The relevant question is not whether that infrastructure will be necessary. It will. The question is how intelligently we build it and who bears its costs.
One promising development is the growing expectation that large data-center operators should provide or pay for the additional power generation and grid infrastructure their facilities require rather than simply passing those costs to residential ratepayers. That creates incentives for new generation—including advanced nuclear reactors, such as small modular reactors (SMR), and renewables.
The AI revolution may therefore force America to confront an energy problem it has postponed for decades.
There is an instructive historical analogy in nuclear power.
Three-Mile Island and Chernobyl became fused in the public imagination as symbols of nuclear catastrophe, even though they were radically different events. Public fear, political incentives, regulatory complexity, and legitimate safety concerns combined to slow the development of American nuclear power for decades. Today, as electricity demand accelerates, the country is reconsidering the very technology it once abandoned.
The lesson is not that regulation is unnecessary.
The lesson is that fear is a poor substitute for discrimination.
We need to distinguish between risks that are possible, risks that are probable, and risks that are merely imaginable. We need to distinguish between technological problems and political problems. And we need to recognize that regulation itself creates consequences, including unintended ones.
There is also a geopolitical dimension that cannot be ignored.
The United States is not developing AI in isolation. China is pursuing advanced AI capabilities as well. Any American regulatory framework therefore has to accomplish two objectives simultaneously: reduce serious risks while preserving the capacity for innovation.
That is not an argument for abandoning safeguards. It is an argument for intelligent safeguards.
A regulatory system so burdensome that it freezes American innovation could create its own national-security risks. A regulatory system so permissive that it ignores genuine dangers would be equally irresponsible. The challenge is precisely the kind of balancing act that ideological politics handles poorly.
But beneath the economics, infrastructure, geopolitics, and regulation lies a deeper problem—one receiving far less attention.
AI does not know the difference between is and ought.
Artificial intelligence operates upon what exists: data, patterns, probabilities, correlations, language, images, behaviors, and accumulated human knowledge. It can become astonishingly proficient at manipulating these materials.
But greater intelligence does not automatically produce greater wisdom.
AI can tell us increasingly sophisticated ways to accomplish an objective. It cannot, by computation alone, tell us whether that objective is worthy.
That distinction is crucial.
Technology has always carried an implicit temptation: if something can be done, eventually someone will try to do it. AI magnifies that temptation because it is not merely a tool. It is a learning tool. Its capacities improve rapidly. It accelerates the distance between what human beings can do and our ability to decide what we should do.
And here we encounter the real AI crisis.
The deepest danger may not be that artificial intelligence becomes alien to humanity. It may be that it becomes too much like us.
AI is trained in human civilization. It absorbs our knowledge, creativity, prejudices, pornography, propaganda, wisdom, violence, compassion, vanity, generosity, and nihilism. In that sense, AI is less an alien intelligence descending upon us than a technological mirror held up before us.
And we may not like what we see.
The ultimate safeguards for artificial intelligence cannot therefore be merely technological. They must be moral and cultural. They depend upon human beings capable of distinguishing between power and wisdom, efficiency and goodness, possibility and responsibility.
We remain capable of shutting down machines.
The more difficult question is whether we retain a sufficiently coherent moral imagination to know when we should.
That is why the AI debate cannot finally be solved by engineers, CEOs, investors, or government regulators alone. It exposes a problem upstream from technology: a civilization increasingly uncertain about the nature of human flourishing and increasingly reluctant to say that some ends are objectively better than others.
AI did not create that problem.
It inherited it from us.
There are genuine dangers associated with artificial intelligence, and prudence requires that we address them. But the greatest danger may be the one least discussed.
We are building machines of extraordinary intelligence at precisely the moment when our culture has become uncertain about what intelligence is for.
The crisis is not simply that AI is becoming more powerful.
It is that the civilization giving it power is increasingly unsure what power ought to serve.
Books by Dr. John Seel, Jr.
0 Comments