Dead Internet Theory: KI übernimmt das Netz | NANO — Transcript
Full transcript
- 0:02In a single day on the internet, 12
- 0:03million fake accounts are removed by
- 0:05Facebook. 50,000 AI-generated songs
- 0:08land on Spotify. More than half of all
- 0:12new Reels on TikTok and Instagram are
- 0:14created by machines. And almost every
- 0:17third new book on Amazon was written
- 0:19with the help of AI. More and more
- 0:22content no longer comes from humans,
- 0:23but from machines. This raises the
- 0:26question: is the internet dead? Is the
- 0:28internet dead? A good question.
- 0:30Every one of us has already been on the
- 0:32internet several times today.
- 0:33I don't think the internet is dead yet,
- 0:35but it’s starting to smell a bit in
- 0:37one or two corners.
- 0:40And this feeling that something isn't
- 0:42quite right on the internet has a name.
- 0:44The Dead Internet Theory. What began
- 0:47almost like a joke on a blog in the
- 0:50late 2010s has developed surprisingly
- 0:52in recent years. But what exactly does
- 0:55the theory actually say? Well,
- 0:59the Dead Internet Theory is, to some
- 1:02extent, a meme that was very popular on
- 1:05Reddit a few years ago and was
- 1:07certainly more of a joke at the time.
- 1:11Yes, it perhaps even started more as a
- 1:14conspiracy theory, but it is a theory
- 1:17that certainly holds much more validity
- 1:19today: namely, that a large part of the
- 1:22so-called traffic, i.e., the websites
- 1:25clicked, is perhaps no longer clicked
- 1:28by humans at all, but by machines, and
- 1:30that the content you see is also no
- 1:33longer created by humans, but by
- 1:35machines. A recent study from April
- 1:402026 showed that automated traffic
- 1:43already accounted for 3%of all web
- 1:46traffic in 2025.
- 1:50There is a concern that eventually only
- 1:52bots will talk to bots, that a computer
- 1:54says: "Hey, so-and-so has a birthday,"
- 1:56automatically writes a greeting, and on
- 1:58the other side, another computer thanks
- 2:00them for this greeting, likes it, and
- 2:01says: "Hey, great." And in the end,
- 2:04there are actually no longer any humans
- 2:06interacting with each other, but just
- 2:07computers. That has nothing to do
- 2:09anymore with the internet that used to
- 2:11exist, where the idea was, hey, we can
- 2:13all participate, the absolute
- 2:15democratization of knowledge, of
- 2:16opinion, of exchange. It was precisely
- 2:20this exchange and the democratization
- 2:22of knowledge that made the internet
- 2:23lively and exciting in the first place.
- 2:27For the first time in history, people
- 2:29from all parts of the world could
- 2:31connect with each other, share
- 2:33information and feelings, and discuss
- 2:35anything and everything. Between 1990
- 2:40and 1999 alone, the number of internet
- 2:42users grew from 2 to 150 million. The
- 2:4890s were the backdrop for the internet
- 2:50revolution, defined by freedom, free
- 2:52knowledge, and cultural exchange. Then,
- 2:56at some point, came the dawn of the
- 2:58blogosphere. People started blogging
- 3:00and exchanging ideas, then came social
- 3:02media, and with social media, there was
- 3:04an internet within the internet.
- 3:06Suddenly, we had a place where we could
- 3:08interact with each other without having
- 3:10to type anything into a browser. I was
- 3:13always on Facebook anyway, for example.
- 3:18With the rise of social media and the
- 3:20spread of smartphones, the exchange of
- 3:22information between people worldwide
- 3:24accelerated at an unimaginable pace.
- 3:29Between 2010 and 2020, the amount of
- 3:31data we generate on the internet every
- 3:34year multiplied by about 30 times. And
- 3:38since the AI boom and the release of
- 3:41ChatGPT in 2022, we are facing the next
- 3:43stage of development. And the next
- 3:46stage will be synthetic content, where
- 3:48basically only computer-generated
- 3:50content exists on the internet, and in
- 3:52the bleakest scenario, it will only be
- 3:54viewed by computers as well.
- 3:57Many bots are already in use today, but
- 3:59what exactly are they? A bot is
- 4:03software that automatically performs
- 4:04tasks on the internet. Incredibly fast,
- 4:07around the clock, and without a human
- 4:09needing to click, type, or watch. Since
- 4:13the AI boom, there are also AI agents
- 4:15and so-called scrapers from companies
- 4:18like OpenAI or Perplexity that scour
- 4:20the web. For example, when a question
- 4:23is asked to ChatGPT, the systems send
- 4:25out small bots. They visit websites,
- 4:27read the content, and retrieve the
- 4:30information so the chatbot can
- 4:31formulate an answer from it.
- 4:35AI-generated content is, of course,
- 4:38flooding social media, and we are
- 4:40finding that social networks can no
- 4:42longer really handle all this
- 4:44AI-generated content. This "slop" that
- 4:47is very, very fast and very, very easy
- 4:49to produce, and of which there is an
- 4:51extreme amount, is something that human
- 4:53creators can no longer compete against.
- 4:57Ömer from Hamburg works as a content
- 4:59creator himself and manages five other
- 5:01influencers at his social media agency.
- 5:05In recent years, he gave up his fashion
- 5:07account, which had over 25,000
- 5:08followers. Among other reasons, due to
- 5:11the pressure from the massive
- 5:12competition of AI-generated videos. I
- 5:19used to make fashion content on another
- 5:21account and then decided for myself
- 5:22that I would give up the channel
- 5:24completely in October 2025. I also
- 5:28noticed that many creators are emerging
- 5:30from the AI sector in that field, which
- 5:32is easier for big brands to collaborate
- 5:35with, and you definitely notice that
- 5:37it's super difficult to get a foot in
- 5:39the door with things like that.
- 5:43AI-generated influencers in the fashion
- 5:45sector are long since a reality. Some
- 5:48reach hundreds of thousands of
- 5:49followers and post several AI-generated
- 5:51videos every day. For many content
- 5:54creators, this means less visibility
- 5:56and the fear of losing their job to AI.
- 5:59Ömer is experiencing exactly this for
- 6:00our industry right now as a manager,
- 6:02seeing the danger that brands will
- 6:03eventually get a taste for it and say,
- 6:05"Hey, I don't have to go through all
- 6:07this trouble anymore." Sending a person
- 6:09, for example, a bag, a coat, a jacket,
- 6:11or trousers so that they can advertise
- 6:13them; instead, I just have to tell an
- 6:16AI: "This is my new pair of trousers,
- 6:18this is the color, I'd like this
- 6:19background, this person, the person
- 6:21needs to be this tall, should have this
- 6:23build." It takes 10 minutes, and the
- 6:26video is created. Basically, such
- 6:28videos are extremely elaborate, and
- 6:30sooner or later, they will naturally be
- 6:33used by these brands.
- 6:35The fact that such content spreads so
- 6:37quickly is not just due to the AI
- 6:39itself. but also due to the platforms '
- 6:41algorithms.
- 6:44The algorithm has a birth defect, I
- 6:46believe. And that is the basic
- 6:48assumption that there is a law of
- 6:49nature whereby content is only
- 6:51interesting to many people if many
- 6:52others have interacted with it. And
- 6:54it's similar with AI-generated content.
- 6:57So the content that I find quite
- 6:59disturbing and strange, I naturally
- 7:01watch, I get stuck on it, and in the
- 7:03end, I give this content the chance for
- 7:05even more people to see it because the
- 7:07algorithm deems it valuable.
- 7:10A vicious cycle that reinforces a
- 7:12central point of the Dead Internet
- 7:14Theory. The internet as a place full of
- 7:17fake accounts, fake comments, and fake
- 7:20reviews, in which humans increasingly
- 7:22play only a passive role. But there is
- 7:26another effect, because the more
- 7:28frequently AI is trained with content
- 7:29that other AI has created, the worse
- 7:31the quality becomes. The results become
- 7:34less accurate, less diverse, and more
- 7:37repetitive. Scientists call this Model
- 7:39Collapse.
- 7:42If we have bots, LMs, systems that
- 7:44create content themselves, which they
- 7:46then use to train themselves again, to
- 7:49create new systems from that, then at
- 7:51some point you have a cycle of
- 7:53distortion that must eventually end at
- 7:55the point where only garbage is
- 7:57produced. And that ultimately means
- 8:00that every day the internet becomes a
- 8:02little worse than it actually once was.
- 8:05And that is a problem for people
- 8:07looking for information. It is becoming
- 8:10increasingly difficult to tell whether
- 8:13a website I land on after using a
- 8:15search engine was created seriously and
- 8:17professionally by humans or if it is
- 8:20AI-generated.
- 8:23Whether what we see, read, or hear
- 8:25online is even real anymore is
- 8:26currently one of the big open questions
- 8:28on the web. Large platforms are trying
- 8:32to counter this with hidden watermarks
- 8:34in files, invisible markings in images,
- 8:36texts, and audio. But the possibility
- 8:40that these solutions could soon be
- 8:42obsolete is real.
- 8:45The dilemma is that, as of now,
- 8:47AI-generated content cannot be
- 8:48technically identified. So, there is no
- 8:51good technology that can detect 100%of
- 8:53AI-generated content. This means we are
- 8:56actually facing a reversal of the
- 8:57burden of proof. We effectively need to
- 8:59define, prove, and verify what is real.
- 9:02I would very urgently like to see
- 9:05editorial obligations introduced.
- 9:07Perhaps only once you have an account
- 9:09yourself with 100, or whatever, 10,000
- 9:12followers. Maybe then you need a legal
- 9:14disclosure, maybe then you have to be a
- 9:16publisher yourself.
- 9:17At the same time, however, we all don't
- 9:18want to give up our freedom on the
- 9:20internet. We want to have the
- 9:21opportunity to be anonymous while
- 9:22online. And I hope that we don't slide
- 9:25into a world where we have to use ID
- 9:27online, because we would lose a great
- 9:29deal that is very important for a free
- 9:32internet.
- 9:35The internet became mainstream in the
- 9:3790s, driven by the idea of an open
- 9:39exchange and freely accessible
- 9:40information. Today, the web is changing
- 9:44noticeably and threatens to become more
- 9:46of a space under the flood of synthetic
- 9:48content where misinformation increases
- 9:50and control grows. So what will the
- 9:54internet look like in 5 years?
- 9:57I believe, or rather, I hope that the
- 9:59stories and the people themselves will
- 10:01endure.
- 10:03I believe that sooner or later we will
- 10:05realize that people have a need to see
- 10:08content from other people. And that
- 10:11will lead to either existing networks
- 10:14finding tools to recognize real content
- 10:17, or new networks emerging that prefer
- 10:19real content in such a way that
- 10:22AI-generated content has a harder time.
- 10:25If we, as viewers, continue to be
- 10:28unwilling to pay for good content,
- 10:31there will likely be two groups. One
- 10:35that continues to get a lot of junk on
- 10:38social media and will no longer find
- 10:41much good content on search engines,
- 10:43and another that is willing and able to
- 10:46pay for good content. The flood of
- 10:52fakes and AI could create a growing
- 10:54divide between those who actively
- 10:56search for real information and even
- 10:58pay for it, and those who only see what
- 11:00is fed to them. This stands in clear
- 11:03contrast to what the internet was
- 11:05originally intended for. Whether this
- 11:09trend can still be stopped is decided
- 11:11by each one of our clicks, because you
- 11:13shape the digital reality of tomorrow.
About this transcript
This page contains the full transcript of Dead Internet Theory: KI übernimmt das Netz | NANO by 3sat NANO, generated from the public captions YouTube serves with the video. The transcript has 1,759 words across 278 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.