pewgar_seemsimandroid to 196 · edit-21 month agomatrix.im ruleimagemessage-square22fedilinkarrow-up181file-text
arrow-up181imagematrix.im rulepewgar_seemsimandroid to 196 · edit-21 month agomessage-square22fedilinkfile-text
minus-squarewholooksherelinkfedilinkarrow-up20·1 month agoYou know asking AI is asking to be lied to yes? Like, Apple has gone so far as to say it’s impossible for current LLMs to reason. It’s incapable of knowing what is true in it’s current form according to apple. Don’t trust AI to actually know anything.
minus-squarewholooksherelinkfedilinkarrow-up2·1 month agoIt’s not that you can trust it a little, it’s that you can’t trust it ever. It’s just saying want it thinks you want to hear, not what is true.
minus-squarepewgar_seemsimandroidOPlinkfedilinkarrow-up1·1 month agoi posted here to see if anyone knew the reason, couldn’t even get a comment on !isitdown@infosec.pub .
minus-squareMidnitte@beehaw.orglinkfedilinkEnglisharrow-up1·30 days agoI mean, by the definition of “LLM”, it’s impossible for the models to reason. It’s literally a fancy big text generation model, like your keyboard text prediction on steroids GPUs.
You know asking AI is asking to be lied to yes?
Like, Apple has gone so far as to say it’s impossible for current LLMs to reason.
It’s incapable of knowing what is true in it’s current form according to apple.
Don’t trust AI to actually know anything.
i didn’t even trust it much
It’s not that you can trust it a little, it’s that you can’t trust it ever. It’s just saying want it thinks you want to hear, not what is true.
i posted here to see if anyone knew the reason, couldn’t even get a comment on !isitdown@infosec.pub .
I mean, by the definition of “LLM”, it’s impossible for the models to reason. It’s literally a fancy big text generation model, like your keyboard text prediction on
steroidsGPUs.