Here’s one for a TV show.

In 2022, a crack quality assurance team was made redundant by a CTO for a botched product launch they didn’t commit.

These men promptly escaped from a maximally unstable job market to the LinkedIn underground.

Today, still wanted by recruiters, they survive as soldiers of fortune.

If you have a broken codebase, if no one else can help, and if you can find them, maybe you can hire the QA-team.

An LLM trained on the PoE ingame chat would try to solve all my problems by asking to buy my Kaom’s Sign Coral Ring for 1ex.

That has happened to me… twice. Once they sent spam to abuse@<domain> and once to postmaster@<domain>. Both of those are “well-known” addresses that received one spam mail each.

Having your own domain with a catch-all address is rare enough that spammers don’t seem to try to target it.

Meanwhile I set up straight-to-spam rules for a handful of companies that leaked my email address. Very useful.

It’s known that AI companies will harvest content without care for its veracity and train LLMs on it. These LLMs will then regurgitate that content as fact.

This isn’t a particularly novel finding but the experiment illustrates it rather well.

The researchers you consider to have acted so immorally did add useless information to the knowledge pool – but it was unadvertised, immediately recognizable useless information that any sane reviewer would’ve flagged. They included subtle clues like thanking someone at Starfleet Academy for letting them use a lab aboard the USS Enterprise. They claimed to have gotten funding from the Sideshow Bob Foundation. Subtle.

By providing this easily traceable nonsense, they were able to turn the generally-but-informally known understanding that LLMs will repeat bullshit into a hard scientific data point that others can build on. Nothing world-changing but still valuable. They basically did what Alan Sokal did.

Instead of worrying about this experiment you should worry about all the misinformation in LLMs that wasn’t provided (and diligently documented) by well-meaning researchers.

Man, Crank 3 is a lot worse than the first two.

Dev: Why was my app rejected?

Apple: Your app was rejected because it uses a payment processor that is not allowed on the App Store. Is there anything else I can help you with?

Dev: My app doesn’t contain any payment functionality at all!

Apple: I’m sorry, I made a like mistake there. Please contact the App Store support team to help you with this kind of issue.

Dev: You are the App Store support team!

Apple: That is entirely correct! Please allow me to refer you to section 14 of the Apple Developer Agreement to show why your payment processor is not supported:

[Link: to the Apple Developer Agreement, section 14: Disclaimer of Liability]

I’d change the second one to “reasonably progressive and actually tech-literate hypercapitalists who, in case they become part of a governing coalition, suddenly oppose everything the government does at the expense of their own credibility”.

Also you forget to mention that the nazis are blatantly abusing their privilege as MPs to spy for Russia.

There’s also the less self-empowered variety where the transferred person not only looks and behaves like an exaggerated Barbie doll but is supposed to become that airheaded. Like so often, the lines between these varieties are blurry.

By the way, I find it rather telling that bimbo fetish is becoming more popular at a time when all of the MAGA women are doing their best to look like mass-produced plastic dolls… Exposure does a lot to drive preferences.

Right. Safe Search doesn’t matter because Google Search has been outclassed for a while now. There’s no drawbacks… other than the fact that apparently OP gets told about this every single day.

The term “nonspell” would be available if the only relevant distinction is whether it’s a spell or not.

Still sounds like a badly chosen name to me. Calling a category “weapon attack” when not all attacks within it are attacks with weapons makes it wide open to misinterpretation, especially when in some cases it’s relevant whether a weapon is used or not. The fact that it took you two long paragraphs to explain the difference between a weapon attack and a weapon attack with a weapon illustrates this rather nicely.

Distinguishing “spell/nonspell” or “spell/weapon/unarmed” would’ve solved the issue without this whole “weapon but not really” song and dance routine.

Melee weapon attacks not being attacks with a weapon sounds like a prime example of badly written rules.

It’s not like you can’t add a “Log in with your company’s SSO” button to the form. That works just fine and at least Microsoft does something like that.

It depends. “Donald Trump wants to annex Wisconsin as the 51st state” wouldn’t be funny. Pocketpair launching a Steam store page for the Palworld dating sim (which was last year’s April Fools joke) was.

But yeah, this isn’t the time for political humor.