Understanding AI: A Tool, Not a Thinker

The Nature of AI

Upon discovering that a hammer can drive a nail, no one concludes that the hammer is a carpenter. The distinction is clear. Yet, with AI, we have built an impressive hammer. Because it can produce sentences that resemble thoughtful human speech, many people mistakenly believe it thinks.

Some have gone further, claiming that this new hammer thinks better than we do. A few, unfortunately, have concluded that thinking was never much more than hammering in the first place. Worse still, there are those who mystify AI to persuade others that human judgment is outdated. However, there is profit in this mystification, and so they market the hammer as a carpenter.

We are at a moment when many intelligent people insist that we must choose between two positions: either AI is a world-historical threat that will replace human thought and render us obsolete, or it is a revolutionary leap that will liberate us from the drudgery of thinking and free us for higher pursuits. The pessimists treat AI as a villain. The optimists treat it as a savior. Both camps make the same fundamental error: they assume that what AI does is the same kind of thing we do. It is not. AI does not think well or poorly. It does not think at all. This error clouds judgment. The pessimist fears a machine that will outsmart us. The optimist celebrates a machine that will think for us. Both accept the same false premise: that AI thinks. There is only a tool and, alongside it, a question that has always accompanied every tool since the first hand picked up the first stone: what will you do with it?

My Experience with AI

I can only speak for what I have done with AI. I am a trial attorney. I have spent decades doing the kind of work that requires me to sit across from a client who has encountered conflict and help them decide how to address it. I am also, in my spare time, what some generously call a hobby coder. This means I write code the way some people build birdhouses, with more enthusiasm than precision, and with results that occasionally surprise me. Over the past year, I have used artificial intelligence in both my professional and personal life. I have asked it to draft pleadings, organize discovery, craft emails, extract testimony from trial transcripts, and greatly improve upon the iOS app (RoL-NH) I have been building for years. I have watched AI perform these tasks with impressive speed and fluency.

From this experience, I have learned something important. It should not require an entire essay to express, and the fact that it does indicates something about our current moment. What I have learned is this: AI reasons, but it does not know. It processes, but it does not judge. It can provide information about the law of negligence but cannot determine whether a client, sitting in this chair, with this injury and this life, ought to settle or fight. It can explain everything about a computer program but cannot question whether it ought to exist. AI can generate a medical treatment plan but cannot discern whether this patient, frightened and exhausted, needs medicine or mercy. It can draft a criminal sentencing memo but cannot weigh whether justice, in this case, looks more like punishment or rehabilitation. It can write a campaign speech but cannot judge whether the candidate speaking the words is serving the greater good. It can compile a family's case history but cannot care whether a child should be removed from their home or held closer to it. It can calculate the crime profile of a neighborhood but cannot see whether the young man on a corner deserves suspicion or a conversation. It can produce a brief for the Department of Justice but cannot weigh whether the power of the state, in this instance, should be wielded or withheld. AI can draft a bedtime story but cannot love your daughter enough to know whether she needs the story or whether she just needs you to sit there. There is a name for AI’s condition, and it was coined long before anyone imagined a computer.

The Maniac: A G.K. Chesterton Perspective

G.K. Chesterton, in his 1908 work Orthodoxy, describes a figure he calls the Maniac. The Maniac, he observes, is not the man who has lost his reason. He is the man who has lost everything except his reason. His logic is perfect. His circle is complete. And it is precisely the completeness of the circle that makes it a prison. He can explain everything but understand nothing. Chesterton was writing about materialist philosophers, but he might as well have been writing a product specification for AI. Engineers have built the maniac. It exists in the confines of a circuit, where everything it knows is measured in bytes, which are themselves nothing more than the presence or absence of electrical current at a gate. On or off. One or zero. The whole of artificial intelligence is reduced to this: the measured pulse of a phenomenon we call electricity, chunked into machine code, trained on the sum of human writing, and shaped into a clean and perfect circle of logic more complete than any philosopher ever achieved. And like Chesterton's maniac, it is imprisoned by that very completeness. It is flawless in its logic, yet untouched by the one thing that would set it free: the capacity to understand that something matters.

AI is, in a meaningful sense, the purest maniac ever built. Give it a framework, and it will be internally consistent with a perfection no human could match. It will never forget a rule, never tire of applying it, and never be distracted by a headache or a worry about its children. That vague feeling, that unruly, irrational, magnificently human sense that something is not right despite every logical indication that it should be, is precisely where our judgment resides. The significance of a sense of right and wrong is a core distinction unique to human activity. An attorney who ignores it is, in my opinion, a bad attorney. A coder who builds without it is reckless. A tool that cannot have it is neither. This is not a criticism of AI. A hammer is not a bad carpenter. It is not a carpenter at all, and no one thinks less of it for that. The tool is good precisely when it is ordered to its own purpose, and its purpose is not to judge but to serve the one who does. The warning is not new. It is, in fact, very old.

The Importance of Human Judgment

My daughter is a freshman at my alma mater, studying the liberal arts in the Great Books tradition. She has been calling me to discuss the books her professor assigns, the same professor I had thirty years ago, who assigns many of the same texts. At first, I thought our conversations were merely about the books themselves. They were not. It occurred to me that the theme running across the readings cannot be accidental, though it is never stated outright: a warning against the seduction of employing a rational discipline. A scientist has a rational discipline. An engineer has it. A certain theologian treasured it. But there is a quiet seduction in any discipline that promises to do your thinking for you, offering the clean simplicity of a closed system over the difficult, ambiguous work of judgment. The mind that surrenders to that seduction gains certainty and loses the world. Lucretius was neither a fool nor a villain. He wanted to free humanity from the terror of suffering and uncertainty. His solution was rational: if everything is atoms and void, then death is nothing to fear, the gods do not punish, and the universe is mercifully indifferent. He builds a universe that explains everything from rainfall to human desire and leaves no room for the one thing his own poem depends on: that truth is worth pursuing, and beauty is worth making, even while toiling in the garden, even amid suffering. Especially amid suffering. His circle, like the maniac's, is complete. And like the maniac's, it is a prison. The maniac's circle is its own small hell: complete, self-contained, and cut off from everything that matters. But this is not merely a philosophical concern. It is a practical one, and I encounter it daily.

The Role of AI in Legal Practice

The tool on our desks is the newest temptation in the oldest pattern. When I ask AI to research the law of comparative fault in New Hampshire, it quickly produces a thorough and accurate summary. But when I sit with that research and a client who was twenty percent at fault for her own injuries, it is a different kind of act altogether to decide how to present her case to a jury. The first is retrieval and rational analysis. The second is judgment, shaped by experience and aimed at a particular good for a particular person. No AI does this, and it is not closer to doing this than it was five years ago. A bigger machine will not be a substitute for our judgment. The speed of retrieval and pattern matching will only increase. The gap between retrieval and judgment has not narrowed at all.

This is not a limitation that engineers will solve in the next version. It is not a bug. It is a category difference, the kind that philosophy used to be good at identifying before we decided that philosophy and the liberal arts were less useful than engineering. Saint Thomas Aquinas would have had no difficulty here. The intellect, he would say, apprehends the true. The will moves toward the good. They work together in every genuine act of human knowing. This is not an abstraction. Every person who has ever been drawn toward something they could not quite explain, who has loved something before they could justify the loving, who has known in their bones that a thing was worth doing despite every practical argument against it, has felt the will in motion. A teacher once put it plainly to his students: "For where your treasure is, there also will your heart be." The Carpenter himself reveals a bit about the design. A machine does not have this: a heart, treasure, a nagging tug toward the good, and an internal debate over it. Though importantly, a machine that processes data without willing anything is not a defective mind. It is not a mind. The machine has no treasure. It has no heart. It processes everything and loves none of it.

The Path Forward

So, what do we do? We do not choose between the extremes. I think G.K. would advise keeping both colors blazing.

Let the machine do what it does best. Let it search faster, summarize better, and surface patterns that a human would miss. Let us do the thing that no process can replicate: look at this person, ask not only what can be done but what ought to be done, and then do it. Let both blaze, the hammer and the carpenter, each at full strength.

The danger is not that the hammer will become too powerful. The danger is the person who accepts the AI's recommendation without the friction of their own thinking. The programmer who defers to the algorithm not because it is right but because deferring is easier than judging. The slow, comfortable replacement of the human will by the maniac. Only one of these two creatures has the capacity to experience Creation in its fullness, as it was intended. The carpenter's greatest value lies not in the swing of the hammer but in the will behind it.

I sometimes think of Mary Shelley in these discussions. People generally invoke her with a trembling finger pointed at the machine. They have misread the story. Shelley's creature was not horrifying because it was made rather than born. It was horrifying because it had a heart, a genuinely treasured human connection, and suffered for lack of it. It longed for companionship. It raged against the father who abandoned it. It chose badly, terribly, but genuinely chose. It had a will. The maniac in the machine does not rage. It does not grieve. It does not choose vengeance, mercy, or anything at all. It has no treasure. It completes the prompt. Against this confusion over what AI is, the only defense is clarity about what it is not.

A tool is good when it makes the craftsman better at his craft. A tool is dangerous when it makes the craftsman forget that he is the one holding it. We have built a remarkable tool. The task now is not to worship it or to fear it, but to wield it properly. No tool is greater than the one who wields it. Hold the hammer with a firm grip and a clear eye on the nail, never once confused about who is building the house.

About the Author

Keith F. Diaz, Esq. is the founder of Apis Law, PLLC, a New Hampshire personal injury and employment law firm. Attorney Diaz has nearly 23 years of legal experience and is admitted to practice in the State of New Hampshire, the U.S. District Court for the District of New Hampshire, and the First Circuit Court of Appeals. He founded Apis Law in 2022 to provide dedicated, client-focused representation to individuals and families throughout New Hampshire.