I write, work, and think about technology far more often than is healthy for a human being. The plus side is that I’m fortunate enough to have a job/hobby that stimulates my brain and is a constant source of inspiration and learning. The downside is awkward silences when I talk about this stuff at parties. This weekend I’m going to try and create a coherent connection between the internet and the AI revolutions.

Why is an AI guy writing about the internet? Well, I’m of the opinion that artificial intelligence is going to be as transformative as the internet. And if that parallel holds and history repeats itself, then unpacking those similarities becomes a worthwhile thing to do on a Sunday night (maybe).

(artwork is by https://superrare.com/infiniteyay#top)

To create a semblance of coherence, here’s the rundown of what I’ll write about:

  1. Introduce the Internet, give a quick history of this massive network, and connect the dots with AI.
  2. Then let’s bring it down to the ground again with the vulnerabilities facing this web of billions of connections.
  3. Whats the “Dead internet theory,” anyone?
  4. I’ll wrap this up, leave the technosphere and take a walk, get some fresh air, and awkwardly reconnect with the biosphere and geosphere.

Let’s start by saying that the internet is truly a wonder of our modern world. It’s up there with the pyramids, great walls, and all that stuff. We have created an immense planetary-scale network, connecting billions of people every day with each other and a vast infrastructure of data centers. Like any good tech-writer, I’m going to throw some statistics around now; unfortunately, these ones are just too big to really wrap our monkey brains around, but hey: “Completionism and stuff.”

  • Almost 70% of the planet’s population uses the Internet.
  • Roughly 33 exabytes (33 billion gigabytes) of data move through the web every single day.
  • The Internet’s estimated data sphere now totals 175 zettabytes (175 trillion gigabytes).
  • Growth isn’t slowing: last year alone, data volume grew by double digits, with even “dead zones” like North Korea starting to log on.
  • Data centers, the Internet’s backbone, consume roughly 1.5% of the planet’s total energy.

Here’s a video from a few years back to help with the visualisation:

If you add all these statistics together and the fact that it’s a technology of connection and communication, it becomes obvious that the internet is not just a tool; it’s a core system of our modern society. A system so pervasive that it affects global politics, economies, culture, and human behaviors.

Handy link about the internet and scale: https://www.wired.com/story/weight-of-the-internet/

How did all of this start? Conspiracy theorists, start your engines, because the internet started in 1966 in the Pentagon. Robert Taylor, then director of the Advanced Research Projects, had three incompatible computers (back in the days when computers took up entire rooms and you had to use a typewriter as an interface). He needed a way to transfer data between the systems. Together with program director Larry Roberts, they sketched the grandfather of the internet: ARPANET on the back of a napkin.

From a napkin to a project to a standard protocol where systems could send packages of data to each other, ArsTechnica takes us through the steps in a fascinating series of articles:
https://arstechnica.com/gadgets/2025/04/a-history-of-the-internet-part-1-an-arpa-dream-takes-form/

Fast forward to 1972 and ARPANET had sent the first-ever email and transmitted aircraft landing data between computers on a private network line.

In 1979 CompuServe was made available to the public. For 5 dollars an hour, people could dial into a private network and receive weather information, share messages, and get financial information.

By the late '80s, there was an infrastructure called the “internet backbone” that was connecting hundreds of thousands of students, teachers, and scientists with each other.

Finally, in 1989, British scientist Tim Berners-Lee at CERN (which was the largest internet node in Europe at the time) created a suite of protocols and concepts that underpin the internet as we know it. He is generally acknowledged as the creator of the World Wide Web as we know it.

These Eureka moments were all the world needed. From this moment on, the WWW was off to the races. The Bush government invested heavily in the backbone of the internet, and the giants of the information age—Microsoft and Apple—jumped on board and pushed this technology into the stratosphere.

If you know the modern history of AI from 2023 to the present day, then the parallels are obvious, particularly the moment of acceleration and the impact of massive government funding from a Republican Party president combined with private sector push from the tech giants. Sam Altman is no Tim Berners-Lee, but perhaps history will view him as such. His company, OpenAI, launched ChatGPT, creating perhaps the AI version of the internet webpage—a simple chat interface where we can interact with a gigantic AI model in a natural and fluent way. Perhaps we could also add the MCP protocol to the list of innovations forming the AI backbone of the next generation of agentic systems.

But returning to the internet, like any successful system, it wasn’t all smooth sailing for the WWW. At multiple points in history, the whole thing almost collapsed. Here are a few lowlights (or highlights, if you genuinely don’t like the internet).

1 Nov 2 1988 — The Morris Worm

  • Background – Cornell graduate student Robert T. Morris accidentally released a self-replicating program that exploited sendmail, finger and RSH bugs. Within hours it had infected an estimated 6–10 % of the roughly 60 000 Internet hosts that existed then. (en.wikipedia.org, alumni.cornell.edu)
  • Mail queues jammed, networks were deliberately unplugged, and the term “worm” hit the front pages; DARPA engineers discussed whether to cut national backbone links to stop propagation. It was the first incident to show that software could bring the entire Internet to its knees. (en.wikipedia.org)

3 1999 - 2000 — The Y2K (Millennium) Bug

  • Background – Software that stored years as two digits risked rolling “00” back to 1900 at midnight UTC on 1 Jan 2000. Governments spent an estimated US $300 B+ remediating code. (asisonline.org, time.com)
  • Media and some analysts predicted simultaneous failures of power grids, ATMs, flight systems and the “collapse of the Internet”. Meticulous patching meant very little actually broke, but the episode remains the archetype of a global digital doomsday scare.

6 Feb 24 2008 — Pakistan YouTube BGP Hijack

  • Background – In trying to censor a video domestically, Pakistan Telecom (AS17557) started an unauthorised announcement of the prefix 208.65.153.0/24. One of Pakistan Telecom's upstream providers, PCCW Global (AS3491) forwarded this announcement to the rest of the Internet, which resulted in the hijacking of YouTube traffic on a global scale.. (ripe.net, wired.com)
  • A single state-level routing error revealed that any ISP could accidentally (or intentionally) reroute a huge slice of the Internet.

9 Oct 21 2016 — Dyn DNS DDoS (Mirai Botnet)

  • Background – Tens of millions of IoT devices running the Mirai malware pummeled DNS provider Dyn, knocking Twitter, Netflix, Spotify and hundreds more domains offline for hours. (en.wikipedia.org, money.cnn.com, wired.com)
  • By targeting DNS — the Internet’s “address book” — attackers showed that a single chokepoint could disrupt vast portions of the web.

10 Dec 2020 — SolarWinds Supply-Chain Attack

  • Background – Russian APT actors inserted malware into SolarWinds Orion updates, giving them access to 18 000+ networks, including multiple U.S. federal agencies. (fortinet.com, zscaler.com, wired.com)
  • Because Orion monitored core infrastructure, CISA warned the compromise posed “grave” danger to the Internet’s trust chain and could be used to alter traffic or software at scale.

I’ll skip all the anchor dragging geopolitic laden moments that make the news these days where a ship “accidently” cuts some massive undersea internet data line, and I’ll also ignore moments where some of the most oppressed people on the planet had the internet switched off at exactly the moment that the molotov cocktails are being thrown.

There’s a new threat though, and its one that comes much closer to my area of expertise.

The Dead Internet theory

In a nutshell this one started as a “the real Paul Mcartney is dead” clunky conspiracy laden theory where everyone with half a brain could figure out that (while it makes a good story) no sorry, its obviously not true.

What is the theory ? Basically that the internet is devoid of human life, its filled with content that has not been generated by a human being and its all being algorithmically curated. It goes several steps further than this, suggesting the human internet died in 2016 with a role for the government and brainwashing etc. But again, for the sake of coherence (and intelligence), I’ll leave it at this. (just walk outside in any city on earth and look at people glued to their smartphones to disprove most of this theory)

Unfortunately, not all of this theory can be disproved though and in 2025 the dead internet theory is turning into something not only plausible, but almost likely to happen. Thanks to Generative AI the internet IS getting overrun by AI generated content and yep, it is being algorithmically curated! In a very strange twist of fate and history, the person (he or she is still anonymous to this day and is only known as IlluminatiPirate) that came up with this idea on a 4chan board in 2021, might have been onto something.

To back up IlluminatiPirate’s theory (I never thought I’d write that, but here we are).

  • The internet is crawling with bots, in terms of pure mass the googlebot is the number one user of the internet by a long shot.
  • Bot traffic now officially accounts for more than half of the activity on the internet.
  • 71% of all webpages in 2025 have AI generated content https://ahrefs.com/blog/what-percentage-of-new-content-is-ai-generated/
  • Social media is a black box for this kind of stuff, but go on any platform and AI content is there and its growing by the minute.

If you just take the top three points and extrapolate what that means for the internet it lines up pretty nicely with the dead internet theory. Bots crawl the internet and the content they encounter is coming from…a bot.

Again just to be very clear, its very obvious that the dead internet theory in its pure form is conspiracy laden nonsense, but it does point to an interesting transformation of modern societies central nervous system. If the amount of AI generated content follows its current trajectory then the internet will become less human and more synthetic, which this raises the question as to whether we will trust it ? In its evolution from the Pentagon, to academic networks, to the internet bubble of the zeros to just being literally everywhere, what is the next step in this gigantic systems evolution? Will it exist only to train the next gargantuan AI model or is AI perhaps the first real pathogen and like any good virus it will create resilience and adaption in its host system ?

Interesting times ahead, curious what your thoughts are on this and thanks for reading,

Kai

All artwork is from https://superrare.com/infiniteyay#top