Is the Dead Internet Theory real?

With the rise of AI language models and the growth of Social Media how much of the internet is really human generated?
Bibliography:
Tweet with no sound:
x.com/historyi...
Tweet of Nintendo Playstation:
x.com/historyi...
Twitter account of historic vids:
/ historyinmemes
Dead Internet Theory wiki:
en.wikipedia.o...
Imperva Bot report:
www.imperva.co...
Vice article about twitter ads being community noted:
www.vice.com/e...
Quartz article on the value of twitter:
qz.com/elon-mu...
Tech Crunch article on IRL app full of bots:
techcrunch.com...

[@chris_php]

Seeing all the bot comments on a video about the dead internet theory is just spot on.

[@SergalCheeseHead]

I enjoy the 6 comments all saying the same thing

[@utsuhoreiuji2309]

I enjoy the 6 comments all saying the same thing

[@GarboSweaty]

I enjoy the 7 comments all saying the same thing

[@cryptospades]

Solid video

[@zephyr8072]

How do I know YOU aren't a bot?

[@chris_php]

If I'm a bot, hopefully I'm a good bot

[@_capunder_]

Yeah theres bot farms and obviously there's bots but almost everyone has Internet

[@zephyr8072]

That's exactly what a bot would say.

[@_capunder_]

@@zephyr8072 mhm coming from the kid with a basic profile

[@zephyr8072]

@@_capunder_ Bot confirmed.

[@_capunder_]

@@zephyr8072 lil bro think calling someone a bot is an insult

[@zephyr8072]

@@_capunder_ That's because your bot algorithms hate the truth.

[@HelloKittySGTC]

enjoy the 8 comments all saying the same thing

[@dropyourself]

Take your meds. Also gpt is prohibitively expensive no one would waste their money with spamming.

[@chris_php]

GPT--4 is only $0.06/1k prompt tokens for 32k context, but also gpt 3.5 is free and there are other free open source LLMs like mistrals which can be downloaded and ran locally.

[@dropyourself]

@@chris_php now scale that up to be able to take a substantial amount of spam which is much cheaper to just do it the old fashioned way of automated responses which costs nothing and is much more consistent for the purpose of spam (which is to pull people to a product or get them on the phone with a human that can do social engineering). Also thinking that a majority of the Internet traffic is spam makes sense, because a majority of UA-cam videos have less than 100 views and are a lot of spam. The point is that the algorithms hide those videos and Twitter is just terrible at that so it looks like a website overrun with bots when, in reality, the bots are just able to cheaply pay to be at the top and aren't separated from the normal content (for example spam ads on UA-cam are separate from the content on UA-cam but the spam tweets on Twitter aren't and only cost as much as a Twitter blue subscription to be placed above all other content)

[@chris_php]

Yes this is why I mention in the video it will be unsustainable in the long run if large amounts of content are generated and interacted with are bots with the lack of ad revenue. But in the future and currently we do see compelling AI generated videos which have voiceovers that are able to grab the attention of viewers particularly on shorts or tiktok. But It is funny seeing the paid checkmark designed to stop bots but it is now encouraging bots.

[@dropyourself]

@@chris_php a tts voice reading out reddit posts over Minecraft parkour will never replace human made content. Generative AI can replace a lot of terrible unoriginal content but that's not altering the internet, all that's doing is making this content easier to produce which will force these platforms to react if they don't want their user experience to be ruined. They already made it so that you have to say if your content is AI generated and, hopefully, they'll add a feature where you can just block all of that content making it easier to avoid that spam (someone will probably just make a chrome extension if UA-cam doesn't do that). Same thing with Twitter where you can block all blue check marks, you might miss out on a little bit of good content but overall it's a good way to avoid spam.
Also the thing with generative AI is that it can't create new ideas, it can create text but that text will either be untrue or stolen by nature of how these LLMs work. They are a tool that can be used to play off of original ideas but they will never be a substitute for original content.

[@chris_php]

Yes I also mention that AI requires a human to train off of and that UA-cam and other platforms will want organic content in their feed. They're already implementing features to mitigate at pure AI content. While LLM are still quite primitive it will be interesting to see how it will evolv over time whether it will plateau or continue to advance greatly.

[@scumbagtrashlord]

I enjoy the 9 comments all saying the same thing