Did bots and AI kill the internet? The Dead Internet Theory

👏 SUPPORT THE CHANNEL:
Get access to:
- a Daily Linux News show
- a weekly patroncast for more personal thoughts
- polls on the next topics I cover,
- your name in the credits

Or, you can donate whatever you want:

👕 GET TLE MERCH

🎙️ LINUX AND OPEN SOURCE NEWS PODCAST:

🏆 FOLLOW ME ELSEWHERE:

#AI #bots #aitechnology #internet

Timecodes:
00:00 Intro
00:51 Sponsor: Squarespace
00:55 Dead Internet Theory
05:38 Bot invasion
09:52 Rise of AI content
12:02 Algorithms: the root of the problem
14:28 Is the Internet actually dead?
16:13 Sponsor: Tuxedo Computers
17:14 Support the channel

The gist of it is that the internet has apparently now been completely taken over by artificial intelligence, and is now completely empty and devoid of real human people, they even called it "entirely sterile". They go as far as saying that all the human generated content, like youtube videos is, in fact, AI generated.

The people subscribing to this theory go even further, they think this is helped by a group of influencers who are working for the government, in order to manipulate our thoughts and get us to buy stuff.

To justify this theory, they use a few things here and there, like the fact that there are so many accounts that just post the same type of stuff. They also say that deepfakes are proof that everything is AI generated.

Bots aren't inherently bad, they can just be tools, like something that displays the score for a sportsball game, or a bot that will save a message for you, or a bot that generates a haiku in a reddit post. But bots are also generally just used to spread crap and scams, and misinformation.
They're here to appeal to people, get a like or two by people who like the profile pic, I guess, and thus build legitimacy for the bot account, so it can then start spamming other videos with scams and URLS.

Which is where we can actually go back to the dead internet theory: yes, some of these bots are created to influence people's opinions on things. Just like TV channels always did, just like real humans always did in podcasts, videos, talk shows, and articles.

Has it gotten worse? yes, it has.

AI isn't just used to power bots, it's also used by bots and humans to post stuff. It is the ultimate form of clickbait, one that takes 0 effort to make, looks real enough to get a quick like from people who are just scrolling with their eyes empty as they sit on their porcelain throne. This is just mimicking things that tend to go viral, but with automation.

You might be wondering: this is really crap content, no one should like this and want to watch it, or engage with it at all. Except, as I said, people who scroll over this in passing tend not to notice, and so people drop a courtesy like, or simply don't report the account. And since all platforms make use of algorithms to show you things they think you'll like, instead of things you're actually following or that you're subscribed to, this crap is shared again and again.

But that's because of money, not because of a government backed, AI driven agenda. If you show people more stuff, they spend more time on your platform, they see more ads, and the platform makes more money. And in the process, influencers with little ethics will notice that, and make their own posts more clickbaity so they can get a bigger share of that revenue.

[davidc]

I am in my mid-fifties and have been using internet protocols since the late 80s. The social media and smartphone era of the internet grosses me out.

[WhatIsItReallyAbout]

Disagree. No bot could be as stupid as some of the comments I've seen.

[mirzaangon]

The number of times I have to tell my mother "it's computer generated, not real" when she shows me something on her phone, is insane

[MenaceInc]

It's true.
Source: I'm a bot

[Technopath]

The tragic part is the internet probably will die as the feedback loop of AI just overwhelms the actual human content. TBH, I only see this ending as a bunch of small isolated communities in a "dark internet" of sorts.

[pimmanders]

I'm not a bot....i hope, not sure anymore....

[elaquen]

At this point effectively every user in a comment section on Twitter with the blue checkmark is a bot, promoting a OF account, a crypto bro or a combination of the three.

[jsCP]

I feel like the real conspiracy is that the person who came up with Dead Internet theory is a time traveler who went back in time to warn people about it. Unfortunately they failed and Dead Internet Theory still became reality.

To be clear, I am joking. At the same time, I don't know how else to explain how dead-on the theory is. It just took some time for it to come to fruition.

[Lampe]

The internet isn't dead yet, but it is at the risk of dying. Large-scale data scraping is nowadays guaranteed to gather a lot of AI-generated content even when trying to filter it out. So for companies trying to train new AIs it's already on the brink of death. For normal humans it's not dead though, as there still is a lot of non-AI content out there (including everything from before when AIs became a thing that is still online) and it's pretty easy to not get overwashed with AI-generated content (yet).
But the more AI-generated content is put into the internet, the more it dilutes the real-human-made content.

[IAmTheSlink]

I wouldn't go so far as to say it the Internet is dead. But honestly, I don't know how anyone can look up anything on Google and not see the bad bot influence.

[MZZenyl]

Lots of the fairly large image-centered subreddits are thoroughly infested by repost bots, especially on subreddits that have dropped in popularity and are therefore not moderated properly. Most of them just copy the image and title directly, while others will try to hide by adding intentional typos in the title, or crop/skew/colorshift the image slightly to make it harder to detect.
I also spotted three bots on the "AskReddit" subreddit that copy-pasting comments directly from the last time a similar question was asked. And I only noticed because I remembered reading one of the original comments. A lot of the image repost bots will have a handful of comments on AskReddit, specifically because it is very active and posts are often repeated, so it is easy for them to go unnoticed.

[tiagotiagot]

It's an old theory; barely plausible back when it was initially proposed, but it is getting extremely more likely as time passes, tech is mostly already here right now.

[dsouth]

Spambots killed usenet newsgroups, too. You could have even a character-only driven network and people will find ways to automate grifting on it.

[VitharPL]

To be honest. Some "biological" Internet users are hard to distinguish from bots (in term of their intelligence being artificial especially), so maybe there is more true to this theory than we think.

[arthurpizza]

It would take a pretty stupid robot to replace me…

[MSThalamus-gjoi]

Right now, there are a million "tells" for AI content, especially visual content (still images and videos) and written text. Compare that to what AI was capable of five years ago, and then project it forward in time five or ten more years. It's going to become harder and harder to tell what's real and what's fake as time goes by. It's already almost impossible to tell whether a well-known voice is real or computer generated. People are already so easily swayed simply by ads that pop up in their Facebook feeds. We're all in trouble when AI-generated video becomes realistic enough to fool the average person, and capable of depicting fictitious events that drive people in the direction intended by their (indirect) creators.

[-_Grind]

The real question is this: "Do the bots use Linux?"

[Taikaru]

The AI is getting so good, it even realizes it's saying just what an AI will say! Linux Experiment's graphics are slightly more advanced than Neuro-sama.

[cateatingbread]

If you ignore social media (and perhaps news) you'd see that the web is very much alivde

[stephanhuebner]

That I understood the "It's just a theory"-reference probably means that I spend too much time on YouTube.