It’s a Large Language Model designed to generate natural-sounding language based on statistical probabilities and patterns - not knowledge or understanding. It doesn’t “lie” and it doesn’t have the capability to explain itself. It just talks.

That speech being coherent is by design; the accuracy of the content is not.

This isn’t the model failing. It’s just being used for something it was never intended for.

What do you even do on Instagram for 16 hours?

I’ll start paying for YT before I ever start watching ads. I won’t as long as I don’t have to, but the second they finally figure this out, I’m subscribing.

It’s well worth the money considering how much I use it and for how long I’ve used it for free. It’s honestly one of the few streaming services actually worth paying for.

Apparently there is a way around it as I’ve been watching YouTube all day and haven’t encountered a single ad.

Well, they don’t in this case either - they just add an extra step to it. You can buy a bit like that off eBay.

I went looking for a screenshot of that scene but got distracted by all the shop-vacs painted as R2D2. Off to buy white and blue paint.

Not to defend BMW here but it’s likely a very specific part that this screw is used for and 99% of home mechanics probably are never going to encounter it. Most likely having something to do with the high-voltage system which you shouldn’t be messing with anyway.

Probably didn’t read it because it was clearly low-quality slop - not because the final output was written by AI.

People don’t mind AI-generated content when they don’t detect it as such. It’s the low-effort garbage they don’t want to deal with.

Nobody has a perfect radar for AI content. This is just the good old toupee fallacy in action: “All toupees look terrible because I’ve never seen a good one” - except the good ones are the ones you never clocked as toupees.