NewsGuard is currently monitoring 3,749 AI-generated content farms across 16 languages. The figure was, as one would expect, smaller last month and will, of course, be greater next month.

The main point is that it has definitely become cheap to publish large amounts of plausible text. The thing that hasn't changed is that the reader decides whether any of it is true; that role has always belonged to the reader rather than to the person or tool (possibly one and the same) doing the writing.

Backstory, the new tool developed by Google's DeepMind, is a good illustration of the field's progress in the right direction, albeit it is not without its problems. It tracks the history of an image on the web over time and identifies signs that the content in question is AI-generated – a task that used to take a fact-checker fifty minutes can now be carried out in about three. Perhaps that is genuine progress. We shall see.

The fact-checking team at India Today, the country's most widely read magazine with an audience of about eight million people, first uses Backstory as an initial editorial check and then individually confirms each result before publishing anything. However, the tool has trouble distinguishing between the faces of two different individuals, and it is limited to what has already been indexed on the web, a point I am certain my readers are aware of, since there can be a considerable time lag between an item being posted and its indexing. A lag that can be significant enough to matter, anyway.

Once again, the tool got faster, but the judgment behind it still belongs to the human.

As I write, though, I wonder which set of rules and judgments, as determined by the various “detectors,” will actually matter in the long run. And, by the way, are they going to tell us exactly which letter and which space and which comma is “AI versus, says, me?” If so, are they willing to die on that hill? Are they willing to sign up for clear culpability in court? If they are not, should we trust them at all? Real questions.

There’s a fine line between a regular guy writing some stories and LinkedIn posts that use AI to edit his work (pointing at me) and the massive content generation farms. And yet, sadly, there appears to be little effort to distinguish between the two. In point of fact, I have written every letter of this post, but I still get “detectors” telling me (and you) that it is AI-generated. These tools are like great big ‘Utilitarian’ hammers, where all the tools see are nails. Perhaps it does some good dealing with the content farm issue, but it doesn’t do the little guy any favors.

Be a nice person. Do your work. Give credit whenever it is due. And always take out the trash.