「黃家駒 Wong Ka Kui」(old account, migrated to Piefed)@sh.itjust.works to Showerthoughts@lemmy.world · 3 个月前Negativity in social media is probably why some people prefer to talk to LLM botsmessage-squaremessage-square29fedilinkarrow-up1116arrow-down16
arrow-up1110arrow-down1message-squareNegativity in social media is probably why some people prefer to talk to LLM bots「黃家駒 Wong Ka Kui」(old account, migrated to Piefed)@sh.itjust.works to Showerthoughts@lemmy.world · 3 个月前message-square29fedilink
minus-squareFishnoodle@lemmy.worldlinkfedilinkarrow-up2·3 个月前Yeah, but the LLMs are trained with data from social media…
minus-square「黃家駒 Wong Ka Kui」(old account, migrated to Piefed)@sh.itjust.worksOPlinkfedilinkarrow-up3·3 个月前But they have a filter tho… If you see some LLMs with a “thought process” thing, you can see that they filter out any that “sounds negative” so it makes the response more pleasing to the end user…
Yeah, but the LLMs are trained with data from social media…
But they have a filter tho…
If you see some LLMs with a “thought process” thing, you can see that they filter out any that “sounds negative” so it makes the response more pleasing to the end user…