floofloof@lemmy.ca to Technology@lemmy.worldEnglish · 4 months agoQualcomm spends millions on marketing as it is found better battery life, not AI features, is driving Copilot+ PC saleswww.tomshardware.comexternal-linkmessage-square102fedilinkarrow-up1441arrow-down18
arrow-up1433arrow-down1external-linkQualcomm spends millions on marketing as it is found better battery life, not AI features, is driving Copilot+ PC saleswww.tomshardware.comfloofloof@lemmy.ca to Technology@lemmy.worldEnglish · 4 months agomessage-square102fedilink
minus-squareTechNerdWizard42@lemmy.worldlinkfedilinkEnglisharrow-up9·4 months agoCopilot sucks. Gemeni is a sassy teenager. Chatgpt 4o is actually halfway decent. When they announced Gemeni had a million context tokens, that was awesome. But it can’t give coherent output to save its life. Useless.
minus-square11111one11111@lemmy.worldcakelinkfedilinkEnglisharrow-up3·4 months ago“Chatgpt 4o is actually halfway decent.” I think I need to redo my parameters cuz you aren’t the first person I’ve seen say this. I wasconvinced it got dumber lol
minus-squareSirDerpy@lemmy.worldlinkfedilinkEnglisharrow-up2arrow-down1·4 months agoIt got worse. We adapted parameters to try to compensate. You didn’t do anything wrong. You’ve just not yet implemented continuous improvement.
minus-square11111one11111@lemmy.worldcakelinkfedilinkEnglisharrow-up1·4 months agoAny recommendations for where to start for improving it’s efficiency?
minus-squarerenzev@lemmy.worldlinkfedilinkEnglisharrow-up2·4 months agoPerplexity.ai is also pretty nice. As far as I understand, it’s just some version of chatgpt but with the ability to search the web.
Copilot sucks. Gemeni is a sassy teenager. Chatgpt 4o is actually halfway decent. When they announced Gemeni had a million context tokens, that was awesome. But it can’t give coherent output to save its life. Useless.
“Chatgpt 4o is actually halfway decent.”
I think I need to redo my parameters cuz you aren’t the first person I’ve seen say this. I wasconvinced it got dumber lol
It got worse. We adapted parameters to try to compensate. You didn’t do anything wrong. You’ve just not yet implemented continuous improvement.
Any recommendations for where to start for improving it’s efficiency?
Perplexity.ai is also pretty nice. As far as I understand, it’s just some version of chatgpt but with the ability to search the web.