Substack now runs AI detection through a company called Pangram, and it puts labels on writers’ work based on what that software guesses. I’ve turned it off for my publication, and I want to tell you why.
First, I’m done with the AI panic. Writers have always used tools. Typewriters, word processors, spell check, editors, research assistants: each one was supposed to ruin real writing, and none of them did. What matters is whether the work is good, whether it’s honest, and whether someone stands behind it. Every word I publish is mine to answer for. I served, I’ve lived these stories, and my name goes on all of it.
Second, I don’t let an algorithm grade my work in front of my readers. Detection software makes probability guesses, and it gets them wrong. Detectors have flagged human writing, including the US Constitution, as machine-made. Other detectors have flagged my own older writing, 100% written by me before any of these tools existed, as AI. I’m not going to let a black box stamp a label on something I wrote and leave you to wonder.
You’re smart enough to judge my writing for yourselves. Read it. If it’s worth your time, stick around. If it isn’t, no label was going to fix that.
Eric