The web is filling with AI ghosts: Welcome to the machine-shaped web

Do you remember the early days of the internet? When it was all about LAN parties for Counterstrike and entire clans running GeoCities pages with neon GIFs and auto-playing soundtracks. Newgrounds and Flash games ate up our after-school hours, while AOL chat rooms, MSN Messenger and Yahoo forums were bustling with strangers making connections.

The so-called “antisocial” kids were busy teaching themselves HTML, PHP or tinkering with early websites. The good ol’ days weren’t polished or corporate; they were messy, raw and unmistakably human.

So, what happened to those days? Broadband happened. Faster internet pulled us out of the basement LAN party vibe and into something bigger. Suddenly, everything could be connected all the time and that’s when platforms started to take over.

The rise of platforms

Anonymity started disappearing with broadband. Social media sites like Friendster, Myspace and Facebook tied a face to the real name and gave birth to the first wave of influencers. YouTube turned everyday people into celebrities, while Wikipedia made knowledge collaboration a reality.

Blogging platforms exploded, online shopping went from niche to normal and Amazon quietly shifted from a bookstore into the everything store. Gamers moved from LAN parties to Xbox Live and PlayStation Network, where entire communities formed around shared worlds.

The internet that once felt like a playground for outsiders became a utility for everyone. It wasn’t just curiosity anymore — it became about connection, commerce and convenience.

Age of algorithms

Growing up, I was always curious about how things worked. At first, algorithms were just methods to solve math problems; step-by-step instructions passed down from great minds. Then we digitized them, turning those rules into code that could simulate physics, break codes and automate tasks.

I still remember my first course in data structures and algorithms, when it finally clicked: The world around us is run by algorithms trying to make sense of how everything works. I didn’t witness it firsthand, but I began to understand how the same logic had played out in the real world: databases had grown massively and efficiency was critical. As data grew, advances in indexing and query optimization shaped how systems kept up.

That was when classics like “Quicksort” became indispensable, because speed was everything. Networking brought routing algorithms that literally connected the world. A lot happened in between, like new indexing methods, breakthroughs in information retrieval and the rise of early search engines. But then came one of the most iconic of all: Google’s PageRank, which changed how we discovered knowledge on the internet.

From there, algorithms stopped being hidden in the background and started shaping what we saw. You could already see it in your messy Facebook feed: posts stopped appearing in reverse order and instead the system learned what you might want to see. Engagement, outrage, clickbait, viral challenges — all of it kept us staring at screens.

People increasingly lived in tailored realities. This was the first glimpse of a web that felt less personal and more machine-shaped. Historically, algorithms solved problems; in this era, they filtered attention and maximized clicks or purchases.

The mobile and monetized era

I still remember getting my first cellphone and the excitement of being connected wherever I was. It wasn’t a smartphone yet, but it felt like a shift. Not long after, phones stopped being just about calls or texts. The iPhone and the App Store turned them into the main gateway to the internet.

That changed the rhythm of how we lived online. Instead of logging in after school or work, the internet started tapping us on the shoulder all day long. Push notifications buzzed, alerts pinged and feeds never seemed to end.

And while that was happening, the business side of the web changed, too. “Free” platforms weren’t really free anymore. Every click and swipe was tracked, bundled into profiles and turned into ads. The internet quietly shifted from being about what we wanted to do to what companies wanted us to see.

Netflix didn’t just give you movies; it decided what you should watch next. Spotify started curating moods. Uber and DoorDash made it normal for algorithms to decide how we move and what we eat. By then, it wasn’t “going online” anymore. You were just always online. But even then, most of what filled our screens was still made by people. That balance began to change as AI took on a bigger role.

The AI era and the dead internet

So how did the AI era really begin? Long before everyone was throwing “AI” into conversation, the groundwork was being laid. We built systems that could recognize patterns, then we fed them endless data to make them smarter. Machine learning was the buzzword, training algorithms to improve as they processed more examples. We taught the machines our language, our preferences, our habits. What we clicked, what we scrolled past, what we lingered on — all of it became training data.

For years, this work was mostly invisible, humming in the background of search engines, ad networks and recommendation systems. But the data kept compounding and the machines kept getting better at reflecting us back to ourselves. By the time AI started showing up in headlines with Google’s DeepMind beating the Go world champion in 2016 and platforms quietly weaving deep learning into everything, the shift was already well underway.

It was also around this time that a strange idea began circulating in forums and fringe corners of the internet: the “Dead Internet Theory.” The claim was that much of what we saw online wasn’t real anymore. It was bots, spam farms, recycled posts and automated filler.

At first it sounded like pure conspiracy, the kind of late-night paranoia you’d dismiss. But spend enough time online in those years and the theory didn’t feel entirely out of place.

Comment sections were flooded with nonsense, Twitter threads padded with bots, YouTube videos gamed by engagement hacks. The internet was still alive, but it already felt less human than before.

But fast forward to today and the evidence is hard to ignore. AI is no longer just curating — it’s generating at scale, flooding the web with text, images and videos that blur the line between real and synthetic.

And even without generative models, the foundation of the theory has shown up in the numbers. Twitch, for example, revealed that nearly half its traffic came from bots, automated accounts gaming views and engagement. Twitter, Instagram and YouTube face the same fight, caught in endless cycles of spam, fake accounts and AI-generated filler. What once felt like a fringe idea has become a daily reality: a large part of the internet isn’t human at all.

Where do we go from here? Do we go back to hanging out in real spaces to escape the noise or do we get comfortable building relationships with AI versions of humans instead?

This article is published as part of the Foundry Expert Contributor Network.
Want to join?