How we eventually find the confidence to declare something human-made is the puzzle to be solved in this era of generative computing.

I was attending the orchestra recently, listening to—of all things, and really—real human people playing music on real instruments borrowed from a video game series, Final Fantasy. It was a lovely couple of hours. But, mentioning all that is also to clarify there were a lot—no, a lot—of geeks in the audience (present company included, of course.)

Three of those geeks sitting behind me were having a conversation that indicated two things very clearly: they worked in IT and they confidently believed that the emdash was a surefire way to tell if a piece of text was written by an AI.

I rolled my eyes. I have words, many of them, that I will someday share about why the emdash is most definitely not a clear tell of AI, but the point I am instead trying to make here is that while we may not yet agree on what those tells are, it is clear that there are a lot of people who will want to know the answer.

I have been puzzling over this as I write here. I think I’m carrying the load but…

Does the spell check on my computer count as an algorithmic assist?

When I post a photo from my phone did I take that photo with skill or with an automated system that made it a better photo?

Are the very algorithms processing the display of this site in a web browser doing work that I should be doing to call this a human made product?

What is the tell?

If I can’t even figure out if I’m making stuff touched as little as possible by AI, how will anyone else? This is a big puzzle.