LibsEatPoop [any]@hexbear.net to chat@hexbear.netEnglish · 11 days agoI hate it when I ask a question in a group and some mofo replies with “according to ChatGPT”.message-squaremessage-square61fedilinkarrow-up1140arrow-down10file-text
arrow-up1140arrow-down1message-squareI hate it when I ask a question in a group and some mofo replies with “according to ChatGPT”.LibsEatPoop [any]@hexbear.net to chat@hexbear.netEnglish · 11 days agomessage-square61fedilinkfile-text
minus-squarejsomae@lemmy.mllinkfedilinkEnglisharrow-up1·edit-210 days agoWhat bothers me is when people post long outputs from LLMs and expect me to actually read it. Seems rude to me. LLMs are wrong around half the time. So there is some value in asking it, depending on the question.
minus-squarecasskaydee [she/her]@hexbear.netlinkfedilinkEnglisharrow-up1·9 days agoThat’s worse odds than a Magic 8 Ball, which at least sometimes admits it doesn’t know
minus-squarejsomae@lemmy.mllinkfedilinkEnglisharrow-up1·edit-29 days agoThe Magic 8 Ball has that failure rate with yes/no questions. LLMs can achieve this rate on open ended questions, which is much more impressive IMO.
What bothers me is when people post long outputs from LLMs and expect me to actually read it. Seems rude to me.
LLMs are wrong around half the time. So there is some value in asking it, depending on the question.
That’s worse odds than a Magic 8 Ball, which at least sometimes admits it doesn’t know
The Magic 8 Ball has that failure rate with yes/no questions. LLMs can achieve this rate on open ended questions, which is much more impressive IMO.