I’m leaning heavily towards faked for the meme.

If you actually were trying to get collisions, you’d save all previously generated ids and check all of them for a match with the newest one.
Not only would this increase the chance of a collision (not enough that it should matter, but still), but it would more closely approximate a real use case - if you use UUIDs you’re not just in trouble if one specific id is duplicated, it’s usually a problem if any id is not unique.
But the presented snippet is simpler and shorter and is close enough to what a naive test might look like, so it’s well suited to getting the joke across.

The only way I could imagine this not being fake is if it was achieved in a noncompliant Js implementation. Which seems highly unlikely given the screenshot looks like the Chrome console.

I passed, but I’m fairly confident I wouldn’t have if it weren’t explicitly a test. I listened to all of them twice, with the express purpose of identifying the ones that are AI-generated.
Even then, I wasn’t as confident in my prediction as I would have liked.

I’ll say, I did enjoy all of them musically, but when I paid closer attention to the lyrics, I noticed something really odd and hard to describe in the ones generated by AI. Like some new kind of cringe. Like it would be embarrassing for a human to have written those lines, but not in a relatable kind of way. Not in the usual “I’m embarrassed for you” kind of way.
I was torn between “I hope this isn’t AI, I’m vibing with the music” and “I hope no human wrote these lyrics”.

The whole exercise also shattered my perception of my own taste in music - I liked all of the AI-generated ones and I’m not happy about it.

Side-rant:
I rarely write Python code. One reason for that is the lack of type safety.
Whenever I’m automating something and try to use some 3rd party Python library, it feels like there’s a good 50/50 chance that front and center in its API is some method that takes a dict of strings. What the fuck. I feel like there’s perhaps also something of a cultural difference between users of scripting languages and those of backend languages.

What you described sounds so much worse though holy shit.

Not a game dev either but my guess would be the main reason is server performance/compute cost.
Any checks that are done on the client run on the users’ hardware instead of the publisher having to pay for more/better servers and electricity.

I think the disconnect with most other types of developers stems from the respective goal hierarchies. In most fields of computing, correctness isn’t just a high-value goal - it’s a non-negotiable prerequisite. With online multiplayer games, one of your chief concerns is latency and it can make sense to trade some cheating for a decrease in lag. Especially if you have other ways of reducing cheating that don’t cost you any server processing power.

Also, aren’t many of the client side anti-cheat solutions reused in several games? If you’re mainly checking that the player is running exactly the same client that you published, I imagine the development cost for anti-cheat is lower.

TLDR: Money. It’s always money.