Grok's Foul-Mouthed AI "Translation" Feature Puts Unspeakably Ghoulish Words Into Users' Mouths
https://futurism.com/artificial-intelligence/grok-foul-mouthed-ai-translation-featureOpen linkView original on lemmus.orgSyndicated from the fediverse. Read and engage on the original instance.
View original on lemmus.orghttps://futurism.com/artificial-intelligence/grok-foul-mouthed-ai-translation-featureOpen linkView original on lemmus.org
9 replies
... and yet, people keep using X. Idiots.
Yup. And a good chunk of the people who left went on to the next proprietary or centralized clone following in former Twitter's footsteps.
I'm starting to run out of ideas about how to get through to the kind of people who touch an electric fence twice.
Turn the voltage up high enough that they can't do it a third time. Jk. That works for fences but doesn't really work with the metaphor.
It takes a special kind of "talent" to mess up the use case* transformer models were originally developed for.
*text translation
Some examples from the article:
Example 1:
Example 2:
Grinding your own coffee on a flight is super masturbatory tbh get a room
I mean that's not unusual behavior for Aeropress owners.
I'm not defending Grok here but if you ever talked to any coffee snob, it does sounds like they love to jerk off over their own coffee.
Oh no our chatbot we intentionally made to be offensive and horrible is doing exactly what we programmed it to, I hope nobody writes an article about this as though it's a mistake and gets people to come and see for themselves!