Post

So now make it a browser extension and set it to block all links with AI content, and we have a killer idea! FB, Instagram and Bluesky should implement it to automatically block AI posts and we will laude them as the saviors of society.

I'm interested in identifying the rates of self-handicapping. Most of the students in my lab also take coding courses and have fessed up to using humanizing software or creating poor/inefficient code simply to minimize the likelihood of detection/

Still works for me: bsky.app/profile/cv-d...

Sharing our results on writing from english language learners below! bsky.app/profile/max....

For pro writers, less of an issue. From the pov of its most-used domain, student writing — kids growing up in this environment without having their skin thickened or language mastered — if the risk of automated false positives influences how they write, that’s a net loss.

I think showing something like "we tested 10,000 books from 2021 and there were zero false positives. And here we tested 10,000 guaranteed human books from after 2022 and again, zero false positives" would go a long way in terms of trust!

Good, they are intentionally weighted toward avoiding false positives which is vital. Sloppers trying to be like "I slopped and it didn't see my slop" as some kind of gotcha is very funny.

Pangram's new model was only just released 7/29/2026. There basically hasn't been any opportunity yet for the failure modes of this AI detector to present themselves. "By all accounts" is an irresponsible claim when there are no real accounts available.

The thing I worry about here is, it’s an arms race: as the LLMs get better, Pangram will get worse, probably even as they retrain it.

> Obviously purveyors of AI writing don't want this. Strong disagree. I'm extremely excited about these technologies, and also think it's a good thing when their application is transparent and detectable. *Plagiarists* don't want this. It's completely orthogonal to supporting AI.

The false accusation concern is real. The “disapprove of AI use in general” suffers from the overly-broad application of the term “AI”: the description of how Pangram use is designed (NB I haven’t read past that yet) is much more like “old school” machine learning than the LLM-driven AI tools.

This is a great question. I'd love to see research on Pangram's accuracy for non-native speakers specifically. Logically its error rate couldn't be *that* high or else its overall error rate couldn't be so miniscule. But any amount of bias makes the tool's use problematic from a fairness standpoint.

It’s enough to justify it. There is no real trade off.

I hate AI slop, but I also tend to believe people are innocent until proven guilty. I can see this ruining some innocent person's career. How do you prove you didn't use AI?

I feel like a big reason that AI detectors got a bad rep is because they are a case of AI guys selling us a solution to a problem they caused.

The naive stylometry that has led to false accusations and witch hunts has left people thinking it's never possible to just tell. Actually, because the concept of prose that did not come from a mind is so new and alien, people have just been looking for the wrong markers.

Are you finding AI submissions that are otherwise publishable? In other words, is AI writing good stuff? If so, who cares if it is AI?