What Is the Wall Street Journal Scared Of?

The Wall Street Journal's opinion section has decided that the people who write for it can use AI and are not required to tell you they did. Their editor, Paul Gigot, called AI a fact of modern life, and said the question that matters is whether the piece is really the author's argument and whether the author has the standing to make it.

I am not going to argue with the first half of that. I use these tools. It says so on my about page, and it has said so since long before this week. Nobody made me put it there.

So the tool is fine. That is settled, as far as I am concerned. What I cannot follow is the second part, where the tool is fine but the sentence telling people about the tool is somehow the problem.

Here is my honest first reaction, and I mean it as a real question and not a cheap one. Does this mean I can delete my disclosure page? Because if the largest business newspaper in the country has looked at this and concluded that readers do not need to be told, then I am maintaining a page that the industry has just decided is optional. I could take twenty minutes tonight and be rid of it.

I am not going to. And the reason is the same reason I put it up.

There is only one argument for keeping quiet about it, and it is this: that telling people would change how they take the work. That is the whole case. Nobody hides a thing that makes no difference. So the decision not to disclose contains, inside it, an admission that the disclosure would matter to somebody — which is the exact definition of the kind of thing you are supposed to disclose.

You cannot hold both ends of that. Either it does not matter, in which case say it and nothing happens, or it does matter, in which case say it because it matters. The one position that does not survive contact with itself is that it matters enough to withhold and not enough to mention.

It is worth saying that the Journal, like most newsrooms, does have internal rules about its own staff using these tools. So the standard is not that nobody needs governing. The standard is that the people on the payroll do, and the outside contributor publishing under the same masthead does not, and the reader looking at the page has no way to tell which one he is reading.

And the scale of it is the part that actually bothers me. I run this after my kids are asleep. On a good day the number of people reading is small enough that I could name most of them. I still thought those people were owed a plain sentence about how the thing in front of them got made. The Journal has millions of readers. Millions. I have to assume some meaningful number of them would want to know, and the paper has decided on their behalf that they would not.

That is the part I would want explained. Not the technology. The math. Why does the obligation to be straight with your readers get lighter as you acquire more of them?

Gigot's test — is this the author's genuine position, does he have the standing to say it — is a good test. It is just a test for whether to publish something. It is not a test for whether to tell people how it was made. He answered a question nobody was asking and left the one everybody was asking on the table.

So no, I am keeping the page. Not because a policy requires it, because no policy requires it. Because I would rather a reader hear it from me on the way in than find it out later and wonder what else I decided they did not need.

That is the whole thing. It costs one sentence.

— Michael