Hacker News Reader

Can we not pretend that AI can meaningfully classify whether something was written by AI?

If it’s Claude - I’d love to see the prompts. This doesn’t read like AI to me. Lots of active voice. Shorter sentences.

Not saying those are signals of human writing but in my experience AI writing is verbose.

26% isn't high and it's concentrated towards the end, so it could be the author just used it to revise some later paragraphs.

26% means not ai 3/4 times...

Would you accuse someone of murder with 26% matching evidence?

26% AI generated? What does that even mean? How is Pangram arriving at that figure?

26% of the sentences tripped the detection threshold for their classifier. That isn't a detection probability on the entire text.

You may be right. But this feeling is mine. Haha.