← Back to context Comment by pona-a 14 hours ago [flagged] 8 comments pona-a Reply idle_zealot 12 hours ago Can we not pretend that AI can meaningfully classify whether something was written by AI? grvdrm 11 hours ago 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. pona-a 3 hours ago 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. physPop 10 hours ago 26% means not ai 3/4 times... loktarogar 10 hours ago Would you accuse someone of murder with 26% matching evidence? tyg13 13 hours ago 26% AI generated? What does that even mean? How is Pangram arriving at that figure? pona-a 3 hours ago 26% of the sentences tripped the detection threshold for their classifier. That isn't a detection probability on the entire text. jdw64 13 hours ago You may be right. But this feeling is mine. Haha.
idle_zealot 12 hours ago Can we not pretend that AI can meaningfully classify whether something was written by AI?
grvdrm 11 hours ago 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. pona-a 3 hours ago 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.
pona-a 3 hours ago 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.
tyg13 13 hours ago 26% AI generated? What does that even mean? How is Pangram arriving at that figure? pona-a 3 hours ago 26% of the sentences tripped the detection threshold for their classifier. That isn't a detection probability on the entire text.
pona-a 3 hours ago 26% of the sentences tripped the detection threshold for their classifier. That isn't a detection probability on the entire text.
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.