Cevilia (she/they/…)@lemmy.blahaj.zone to Fuck AI@lemmy.worldEnglish · 11 hours agoHallucination vs realitymedia.piefed.socialexternal-linkmessage-square28fedilinkarrow-up1115arrow-down12file-textcross-posted to: onehundredninetysix@lemmy.blahaj.zone
arrow-up1113arrow-down1external-linkHallucination vs realitymedia.piefed.socialCevilia (she/they/…)@lemmy.blahaj.zone to Fuck AI@lemmy.worldEnglish · 11 hours agomessage-square28fedilinkfile-textcross-posted to: onehundredninetysix@lemmy.blahaj.zone
minus-squareMCasq_qsaCJ_234@lemmy.ziplinkfedilinkEnglisharrow-up1·edit-29 hours agoAccording to data from Metr, AI has been improving in its effectiveness at completing long tasks. Here we see the tasks that equal or exceed 50% success. On the other hand, we see tasks that equal or exceed 80% success. The trend may continue along these lines in the coming years, although there is a possibility that it will not. However, AI still has a long way to go before it can match the 8-hour workday in the United States if we count the 50%. But if we talk about 80%, it still has a long way to go.
According to data from Metr, AI has been improving in its effectiveness at completing long tasks.
Here we see the tasks that equal or exceed 50% success.
On the other hand, we see tasks that equal or exceed 80% success.
The trend may continue along these lines in the coming years, although there is a possibility that it will not.
However, AI still has a long way to go before it can match the 8-hour workday in the United States if we count the 50%.
But if we talk about 80%, it still has a long way to go.