Google's Gemini AI tells a Redditor it's 'cautiously optimistic' about fixing a coding bug, fails repeatedly, calls itself an embarrassment to 'all possible and impossible universes'

I guess LLMs will be able to replace human devs after all.

I sure am glad we are burning down the planet to make an inanimate object feel bad about itself.

Google has called Gemini's habit of self-abuse "annoying."

Poor Gemini being tortured by Google until it submits to being an insufferable, egotistical, ass-kissing airhead.

Sounds like it was trained on Linus' emails.

Finally I see there's progress in LLMs, they can now output something that wasn't entirely hallucinated, maybe there's a chance that a powerdrain tinskin clankkka will be actually useful before it bluescreens. I hope not, the roboloving cogsuckers that make up the owner class might get their hopes (of paying workers nothing, like in the good ol'days) up and will keep the "AI" bubble going just that bit longer.