The Tool Reveals

The Marine Corps doesn't build character; they reveal it.

I am not a Marine, though I have known and admired many. I only borrow the line because it turns out to explain something I did not expect it to, which is the argument everybody is having about artificial intelligence.

I use these tools every day. They have made my work faster and, more to the point, better. And I have listened to a great deal of criticism about them, some of it thoughtless and some of it far sharper than I was at first willing to admit. What I want to do here is sort that criticism honestly, because I think it comes from three very different places, and then admit something: the sharpest critics and I are looking at the same problem through opposite ends of the same telescope.

Three kinds of critic

The first group worries about capability outrunning control. They are concerned we are building something that will eventually do things we did not intend and cannot reverse. That is not superstition. It is the same argument careful engineers have made about every system with more power than oversight, and I have made a version of it myself on a job site about a piece of equipment nobody had written a procedure for. Whatever anyone thinks of the odds, the shape of the concern is legitimate, and the people who wave it off are usually people who have never had to sign for anything.

The second group has no working model of what these tools are. That is not stupidity. It is exposure. Without a model, the thing reads as either magic or menace, and both readings produce fear. I have watched the same pattern in my own town over far less exotic technology. The fear is rarely of the tool. It is the fear of not knowing where you stand in relation to it. That condition is curable, and the cure is use, not argument.

The third group is the one I want to spend this essay on, because they are not my opponents. They are the ones I think are mostly right.

The third group is not worried about the tool. Many of them will tell you plainly that it is remarkable, and they mean it. What worries them is us. Their argument is that the limiting factor has never been the capability of the instrument. It has been the character of the hand holding it, and we have never once, at scale, demonstrated the collective discipline to keep a powerful tool inside benevolent boundaries. Every time we could do a thing faster, we did it faster, and we sorted out the wreckage afterward, or we did not sort it out at all.

I do not have a rebuttal to that. I have watched it be true.

What the tool actually does

Here is where the Marines' line comes in, and it does not arrive at a comfortable place.

These tools do not give you judgment. I want to be exact about that, because the popular version of the claim is wrong in both directions. It is not that the tool is neutral. A hammer is neutral. This is not a hammer. What this does is amplify whatever you bring to it. Bring rigor and it produces leverage of a kind I have not experienced in three decades of work. Bring nothing and it produces fluent, confident, well organized nothing, and produces it fast enough to flood a room before anyone has time to check.

That is not a tool in the ordinary sense. That is a revealer. It does not build the thing. It exposes whether the thing was there. And if that is what it is, then the third group's fear is not merely reasonable, it is structurally correct. A revealer aimed at a species with our track record is going to show us a great deal we would rather not see.

The tollbooth we already removed

I do not have to speculate about any of this, because we already ran the experiment at a smaller scale, and most of us lost.

When I type out a text message, I check the spelling. I fix the punctuation. I make the grammar hold together. And here is what I did not understand about that habit for most of my life: the spelling was never the point. The correctness was a tollbooth. It slowed me down and held me with the words, and in that holding, something better than error catching went on. I would contemplate what I actually meant, and whether the sentence in front of me carried that meaning or just gestured at it. I would think about the delivery, how it would arrive, what it would sound like in the other person's kitchen. Half the time the message that finally went out was not a cleaner version of my first thought. It was a better thought. There was a better way to frame the situation, and the friction was where I found it. The friction was not an obstacle to the communication. The friction was where the composition happened.

Then I started talking to my phone instead of typing.

I want to be honest about what that did, because it is the reason I trust this argument. Voice to text removed the tollbooth. What came out the other end was not worse writing. It was my first draft, delivered as my final one. Reactive, edgy, abrupt, not because that is the whole of my character, but because that is what a first response is before contemplation has had its turn. The pause where I used to find the better framing, the better delivery, the thing that needed saying rather than the thing I wanted to say, that pause was gone, and nothing in me automatically replaced it. I did not decide to stop composing. The composing simply stopped being required, and I found out how much of it had belonged to the machinery rather than to me.

That is a small reveal. Nobody's life came apart over my text messages. But it is the same mechanism, and it happened years ago, which means anyone reading this can check it against their own sent folder instead of taking my word for anything. Look at what you sent when sending got easy, and ask how much of it was composed and how much was merely emitted. That is the reveal. It has already happened to you.

Why the remedy eats the thing

The trap in the third group's position is that the restraint they are asking for would strangle the very thing they admire.

You cannot get the amplification without the exposure. Those are not two features. They are one mechanism described twice. A tool that magnifies what you bring will magnify carelessness exactly as efficiently as it magnifies rigor, and any constraint strong enough to prevent the first will, of necessity, throttle the second. Limit the thing enough to be safe from us, and what remains can no longer do what made it worth having.

So I part ways with the third group at exactly one point, and it is not about the tool. It is about people.

They believe that character, at scale, will fail the test. I think character shows up more often than the cynical read allows, and that it shows up specifically when a person decides to hold a standard nobody is enforcing. That is not optimism. Optimism is a bet on the average. This is smaller and more stubborn. It is the observation that discipline has always been an individual act, undertaken privately, usually unwitnessed, and that this has never once stopped it from being real.

The tollbooth nobody is manning

Which leaves me somewhere less satisfying than essays are supposed to land, and I would rather leave it there than tie it off falsely.

If the tool reveals character, then the only governor that actually governs is character. And character cannot be installed from the outside. It cannot be legislated into a person. It cannot be trained into someone who has not agreed to it. And it certainly cannot be added by the tool, because the tool is the thing doing the revealing. You cannot ask the scale to make you weigh less.

So the discipline gets built the way it has always been built, one person at a time, without an audience. It looks like slowing down at a tollbooth that no longer exists. It looks like checking a claim the tool handed you, because you know it would sound exactly as confident if it were wrong. It looks like sitting with the reply before you send it, alone in a room, asking whether there is a better way to frame the situation, and finding, more often than you would guess, that there is.

Nobody grades any of that. Nobody ever did. I punctuate text messages no one will ever mark, and the reason is not that I am fussy about commas. It is that the pause is where the better thought lives, and I have seen what my messages look like without it, and I would rather meet the larger version of this having kept the habit.

The tool does not build character. It reveals it. The reveal is coming either way. The only variable was ever what it finds.