[jerf]

If blogging is dead because of AI, videos are not far behind. I'm already curating away a couple channels a week on YouTube that are pure AI spam, either videos that just steal content from others and use AI to drive a collection of clips (which will only get more sophisticated) or put a TTS-driven monologue over something, where the Text being converted To Speech might as well be AI-driven.

And the human videos are already so optimized for The Algorithm that a good deal of them might as well just cut out the middleman and be written by an AI, for the AI. ("This Breaks the Internet - Reacting to Reaction Videos Reacting To Reaction Videos". You can see the thumbnail already, can't you?)

Yeah, they're behind for now, but all the infrastructure is there for this to explode the instant they're not. Video operations putting out dozens of crap videos a week with nonfunctional "hacks" and all sorts of similar things will be in an arms race to pivot to these things and there's no chance the video site's moderation will be able to keep up with this in the end.

[pjc50]

"Dead Internet Theory".

Nobody has to delete the authentic content, they just have to outcompete it by many orders of magnitude. Then not only is the inauthentic content all over all the possible routes you might find content, people stop looking for it entirely.

[masterof0]

Videos are harder to fake and produce, I would say impossible, any of the OpenAI products are unable to be creative.

[jerf]

If you are a YouTube user and have curated your experience as I have, open a browser incognito window and hit the home page of YouTube fresh.

"Creative" as a criterion for video success is at the very least under assault. Looking at what I get, I am not quite so cynical as to say it's dead. It's not. But it is certainly under assault.

There is also the problem of a small number of truly creative content creators getting their content and ideas yanked by people armed with AIs who can then completely out compete the actual creatives.

This has, to an extent, actually already happened for very young child content. Look at "Elsagate": https://en.wikipedia.org/wiki/Elsagate but for the purposes of my comment here, ignore the disturbing nature of the content and instead observe that the content obviously had heavy computer generation influence, and could only be cranked out more quickly and effectively with some rather simple AI.

Like I said, it's not here yet, but the first derivative is clearly positive here, and the second probably is too. Pure AI content that looks like live action is not a prerequisite.

I'm not entirely sure Elsagate itself didn't itself source from some AI in some critical manner. It doesn't seem to have completely created the video from top to bottom, but it looks an awful lot to me like what you'd expect an AI writing the scripts, with the feedback provided by view numbers, and completely unscrupulous humans implementing the scripts for the views, spinning off into a hyperoptimized regime for that one goal at the expense of all else. And, you know, it basically worked. If anything it worked too well. The first AI corruption of video sites may well be in the past, not the future.

[joebob42]

I've never understood the elsagate content. Why are they making it? Is it the sort of thing that young children find appealing if left alone? I can't understand why that would be, and it doesn't seem substantially easier to make than the equivalent content without the disturbing/ inappropriate edge.