David Gerard@awful.systemsM to TechTakes@awful.systemsEnglish · 7 months agoRemember how ChatGPT totally aced the bar exam? Wow! yeah, turns out that was just a liewww.nytimes.comexternal-linkmessage-square172fedilinkarrow-up1639file-textcross-posted to: fuck_ai@lemmy.world
arrow-up1639external-linkRemember how ChatGPT totally aced the bar exam? Wow! yeah, turns out that was just a liewww.nytimes.comDavid Gerard@awful.systemsM to TechTakes@awful.systemsEnglish · 7 months agomessage-square172fedilinkfile-textcross-posted to: fuck_ai@lemmy.world
minus-squarevrighter@discuss.tchncs.delinkfedilinkEnglisharrow-up7·7 months agoeven if that wasn’t the case, a 90% success rate is absolutely abysmal in practice.
minus-squareCouldbealeotard@lemmy.worldlinkfedilinkEnglisharrow-up46·7 months ago90th percentile means it performed equal or better than 90% of the comparisons, no? Not that it got 90% score.
even if that wasn’t the case, a 90% success rate is absolutely abysmal in practice.
90th percentile means it performed equal or better than 90% of the comparisons, no? Not that it got 90% score.