They’re already known to be incredibly persuasive, to the point where some LLMs are made to refuse requests for persuasive political content. They don’t even need to be manipulative—but I suspect a segment of scammers will be more than happy to do this. LLMs are an infinite attack surface.

I know we all inhabit history but I would prefer to be living in an era that gets a few lines in the book and not multiple chapters

I pay @discord.com money for their buggy apps and lack of engineering contact points and they still shove ads at me (Quests). If I give you money for your product, you don't get my time, too.

It's a crime modern XML tooling is scarce and XSLT 3.0 is practically only supported by Saxon. You can't debug in the browser either, they halt on error with no output and they're stuck on older standards.

REMEMBER: All the "Don't use em-dashes" "don't use rule of three" advice because it makes your writing "look like AI" is because they stole actual writers work, and their "predict the next thing" stuff just simulates what WE FUCKING DO. Keep writing well. Use every style and trick you have. Fuck AI.

Must have been written by Copilot.

If BrandShield didn’t want folks to call it a “trash AI-powered” product, maybe they shouldn’t have built a trash AI-powered product.

(BrandShield asked Gizmodo to remove ‘trash’ from the headline of its article about their “trash AI-powered” product’s false claims against @itch.io.)

I had an idea for an episode of Star Trek: Voyager. I swipe-typed “Neelix”. I got “Netflix”.

So Janeway is asking Chakotay to take Netflix and reconnoiter the area.