I’d like to be able to get a rating of whether an article I’m reading is likely to be LLM generated or not, as a measure of how much I should trust it. Ideally I’d like this in my browser as an extension alongside UBlock and consent-o-matic.
Does anyone know if such a thing exists? I found Winston from a quick review of the extension store but it is a paid extension and I’d rather something free.
From what little I’ve read there are many organizations claiming they can detect AI written content (I don’t know about plugins) but there’s little evidence that they’re accurately able to do so.
Tools like Turnitin or GPTzero don’t work well enough to trust. The real issue isn’t just detecting AI writing. It’s doing it without falsely accusing students. Even a 0.5% false positive rate is too high when someone’s academic future is on the line. I’m more concerned about wrongly flagging human-written work than missing AI use. These tools can’t explain why they suspect AI. At best, they only catch obvious cases. Ones you’d likely notice yourself anyway.
The LLM grifters have, indeed, spawned a spin-off community of grifters targeting the anti-LLM community.
It’s grifters all the way down.