Pangram sucks as much as humans who think they're perfect llm generated text detectors, too.

As qarl2 wrote below:

>See how perverse this "I hate the way AI sounds" thing is?

>In the end everyone accusing everyone of being AI.

Which — as we all know — was trained on how humans write.

If it's not one thing, it's another...

https://www.youtube.com/watch?v=k59d-xMvooA&t=205s

Please give one example of a long-form verifiable non-adversarial pangram false positive (e.g., something written before 2022 that gives a result other than 100% human.)

I've never seen a false positive from Pangram for human-written text of 300+ words.

Seems like you have now?

https://news.ycombinator.com/item?id=49372618

People sometimes tell lies on the internet. This is definitely AI-generated.

Pangram sucks by seeding fear.

Say more? I’ve found it quite useful and not a source of fear. It’s effective at calling out disrespect (human to human communication where one party stops participating as a human), and has been effective for me in writing my own communications more crisply and carefully

Fear of being tagged as AI even if you write the text yourself.

For me personally, it's fear to use – and —. I have my versions of XKB keyboard layouts that handily map these but I don't use them anymore. Nor do I use emoji, because some 18-6 months ago AU output abused emoji.

I honestly don’t think those are going to get you flagged as mostly-AI or with high confidence. On anything longer than a sentence I’ve found the false positive rate to be perfectly reasonable.