I am the AI

I’ve mentioned this a few times on social media over the last three or four years or so as the LLMs have come to the fore. Whenever I’ve checked my own writing against an AI that’s supposed to detect LLM use, they usually come back with a high percentage chance that my writing was AI generated. It wasn’t. It isn’t. Some of the material I checked was from the 90s and they still claim that it’s AI… …the obvious explanation is that the LLMs have been trained on my work, they have assimilated my writing style and so when checked a piece of my prose somehow resembles what an LLM would output. I am the AI! Along with dozens of my colleagues on the hundreds, if not thousands, of outlets we all write for. Now, I recently pointed out that all of these AIs that check for LLM use will have false negatives and false positives. Someone replied, giving me the name of one such tool that they claimed was 100% accurate. Nothing can be. But, I gave them the benefit of the doubt and tested it with some…