I'm surprised that I hadn't recognized what dQw4w9WgXcQ means by now. I wonder how many people do.

I didn't realize the source, but when you posted I was pretty sure I'd seen it elsewhere:

https://news.ycombinator.com/threads?id=dQw4w9WgXcQ

You guys are awful, you know that? Discussing this URL without spoilers... it's because of people like you that that thing has so many views!

:P

I recognized it immediately when I was watching the video :^)

When it showed up I sort of guessed, but had to try clicking it anyway, then my wife asked why I was laughing.

The video subsequently shows the source submission: https://news.ycombinator.com/item?id=27995270

which seems to be the most popular submission with a YouTube URL in the past month.

HN search seems to prioritize text matches before the URL matches when I search for "https://www.youtube.com", but the first URL match is for that submission.

I suspected what it must be when my browser autocompleted it... this isn't my first time visiting that special place.

Codex is never gonna let you down.

So the question is whether this is real or just a troll.

It's real. I was totally surprised when that was the URL it picked.

You asked it to do something with "the video on youtube" but what does "the video" refer to? It seems the most likely url associated with the phrase "the video on youtube" is, well, that.

So basically it failed at anaphora resolution.

Seen another way, you asked it for "the video" and so it gave you the video.

I made a similar little GPT-3 toy (for Linux/bash) that also ended up generating a Rickroll URL.[0]

In vague cases, GPT-3 functions approximately as well as a markov chain - it will give you the most “probable” sequence of tokens. There’s some implementation details regarding how it deals with tokenization, but unless you really increase the randomness (“temperature”) you’re likely to get very popular YT video IDs.

In other cases, it will “hallucinate” things like URLs, UUIDs, hashes, and other things that are basically a random string of characters. In my experience it will make UUIDs that have the right number of characters but seem suspiciously non-random and don’t fit any of the defined UUID formats.

Fun stuff.

[0] https://youtu.be/j0UnS3jHhAA

> it will make UUIDs that have the right number of characters but seem suspiciously non-random and don’t fit any of the defined UUID formats

As predicted by Scott Adams in October 2001 ;)

https://dilbert.com/strip/2001-10-25

Expected a Rickroll but got an ad for a mobile game.

Heh