YouTube used to have Gmail-style auto-replies like “Thank you!” and “Glad to hear it.”

I’ve never used them, but I DEFINITELY won’t now because they’ve updated them to be
A) longer
B) generated by AI

Like, look how bad this is!
“We’ve set personal boundaries, maybe we’ll stick to them, tee hee!”

@Graham_LRR waits for the ai to start adding in a random uwu

@Graham_LRR why is the youtube comment ai flirty? did it get confused with a dating app?

are they trying to automate catfishing.

@Graham_LRR Lex did specify that the boundaries part was an IRL thing so I guess the AI took it's chances :/
@struki @Graham_LRR It took a calculated risk, but man, are LLMs bad at math.
@LexYeen @Graham_LRR tell me about it - they can't even manage "how many fingers am I holding up"
@Graham_LRR Hooooooooobuddy. I'm glad I've never even seen that feature.
@Graham_LRR that's .. spectacularly bad.
@Graham_LRR Obviously, the main selling point of AI is removing accountability. Having it not understand what boundaries are is basically built into its DNA.
@Graham_LRR Wow, this new example of just now dumb LLMs are is...a lot.
@Graham_LRR is there a way to completely turn them off?

@Graham_LRR If corporations had any level of restraint, they maybe would have realized that "AI" is not ketchup software.

But nah, they gotta put it on EVERYthing, just in case they stumble upon the thing that will make them ALL the money.

@Graham_LRR I'm starting to feel that "upsetting output" is an intentional feature of genAI.
I mean really it's is that it's being pushed for by overconfident nincompoops who don't want to do any kind of testing before rolling features out.
But dear lord has "offputting" become the easiest identifyiable feature of genAI.