Elon Musk’s Grok keeps making up fake news based on X users’ jokes
istanbullu, I don’t understand why people act suprised about this. These models mimic the data they are trained on. Of course they behave in this way.
If you train a language model on Twitter data, then you will get a model that sounds like a Twitter user.
DontMakeMoreBabies, Curious what sort of prompt they fed it.
db2, “Imitate 3% of Elon Musk’s insanity.”
Hiko0, Just like him. It‘s a match made in heaven hell.
tsonfeir, 
wtf is grok?
db2, It’s Elon being an unoriginal turd again.
magnetosphere, 
I had never heard of it either, but once I learned Musk was involved, the section title “Grok ranks among most reckless chatbots” made perfect sense.
Dearth, (edited ) Grok is an old ttrpg term that means “to intuitively understand something.” As in “I totally grok how to play as an elf in ad&d but playing as a wizard is totally over my head”
edit: correct the autocorrect
tsonfeir,
surewhynotlem, “to really understand something”
Eldritch, It’s also an old Log parser and search function for Linux/unix. So it’s logical he would steal it like he stole the xorg iconography.
Interstellar_1, 
It’s X’s integrated ChatGPT AI
coffeebiscuit, Elons version of being artificially smart.
ObviouslyNotBanana, 
It’s ChatGPT but apparently entirely about X jokes? Lol
clay_pidgin, It comes from the Robert Heinlein sci-fi novel Stranger in a Strange Land. It was defined as fully understanding something. It was used later by nerds in other contexts, as your other replies show.
Add comment