Would you still want to be online if everyone else was a bot?

A few years ago, the Dead Internet Theory was one of those wonderfully strange internet conspiracy theories that was interesting to read about but easy enough to dismiss. The theory suggested that at some point bots and automated systems would make up the majority of activity online, while actual humans would gradually become a smaller part of the internet they had created.

It sounded suitably dystopian. Except lately, parts of it don't sound quite so ridiculous.

Cloudflare CEO Matthew Prince recently shared a graph showing automated traffic passing the 50% mark, with the comment, “Well, that happened faster than I predicted.” There are some important caveats to that number, because bots aren't new and not all automated internet traffic is remotely sinister. Search engines, monitoring services and all sorts of perfectly useful systems have been crawling around the internet for decades.

What has changed is what the machines are doing while they're here.

For most of the internet's history, automated systems largely moved around an internet made by people. Now we have machines creating enormous amounts of the internet as well. Generative AI can write the article, make the photograph, create the video, publish the social media post, leave a comment underneath it and respond to somebody else's comment. AI agents are beginning to browse and interact with websites on our behalf, and this is only going to become more sophisticated.

That creates a rather strange question for those of us who actually live a reasonable part of our lives online: at what point do we stop knowing whether we're interacting with other people?

Something about the internet feels different

I don't think you need to know anything about the Dead Internet Theory to notice that online spaces have changed. We're becoming accustomed to looking at things with a degree of suspicion that wasn't there before.

A remarkable photograph appears in your feed and you look at the hands, the background or the lighting to work out whether it's AI. A video seems extraordinary and your first reaction is no longer necessarily amazement, but doubt. You read an overly polished LinkedIn post from a senior executive and wonder whether they actually wrote any of it. Product reviews, comments, profile pictures, articles and even supposedly candid photographs now come with the same small question attached: is this real?

None of those things on their own represents the death of the internet. Nor do I think using AI to help create something automatically makes it worthless. I use AI myself, and there is a huge difference between using AI as a tool to help develop, challenge or communicate your thinking and generating endless content simply because you can.

The problem is what happens when synthetic content becomes so widespread that we start losing confidence in the human layer underneath it.

The internet has always contained things that weren't what they appeared to be. We've had fake reviews, spam, bots, scams, manipulated photographs and people pretending to be somebody else for as long as we've been online. But there was still a fairly reasonable assumption that most of the spaces we occupied were fundamentally populated by people.

Generative AI starts to change that assumption.

What happens when we stop trusting the internet?

Trust is one of those invisible pieces of infrastructure that we don't tend to notice until it begins to disappear. If I believe the person recommending a product has actually used it, their recommendation has value. If I believe someone sharing an experience genuinely lived through it, I engage with their story differently. If I join a discussion because I want to hear what other people think, it matters whether those opinions actually belong to people.

Once we're uncertain about those things, our behaviour changes.

We become more sceptical about what we read and see. We start looking for signals that something is authentic. We retreat towards people and communities we already know. Eventually, some people may decide that participating isn't worth the effort at all.

That, to me, is much more interesting than the debate about AI slop. The real issue isn't simply whether AI-generated content is good or bad. It's whether an internet saturated with synthetic activity can maintain enough trust for humans to continue wanting to participate in it.

I don't think abandoning the internet is the answer

There are people who are already pulling away from large public platforms, while others are gravitating towards smaller communities, newsletters, private groups and online spaces where they have a better sense of who they're actually talking to. There are also people arguing that we need to rethink the architecture of the internet itself, decentralising it and creating spaces that aren't dominated by a handful of enormous platforms and their algorithms.

I understand the temptation to leave, particularly when scrolling through yet another feed full of recycled posts, synthetic images and content apparently designed for no purpose other than feeding an algorithm.

But I don't want to give up on the internet.

For all its faults, the internet remains one of the most extraordinary pieces of shared infrastructure humanity has ever created. It allows people to find communities they could never have found locally, gives us access to an extraordinary amount of knowledge and enables ideas to travel across geographic, cultural and economic boundaries at remarkable speed. It has allowed people who were previously excluded from traditional publishing, media and institutions to find an audience of their own.

That is worth preserving, even if the internet we preserve looks quite different from the one we've known.

Perhaps AI changes what we value online

For a long time, the economics of the internet have rewarded volume. Platforms want us to publish frequently, generate engagement, attract followers and keep producing because there is always another feed that needs filling.

AI is spectacularly good at volume. It can produce more articles, images, videos, comments and social media posts than humans could ever hope to consume. If volume remains the thing we value, machines will win that competition easily.

But perhaps that's exactly what changes next.

When content becomes effectively infinite, the scarce thing isn't content anymore. It is trust. It is knowing where an idea came from, understanding the experience behind it and having some confidence that there is a person whose judgement, expertise or perspective you value.

That could make smaller communities more important. It could make trusted writers and creators more valuable. It may mean we become more interested in lived experience and original thinking than perfectly polished content. We might even become more tolerant of the imperfections that used to bother us, because those imperfections become useful signals that somebody was actually there.

There is a lovely irony in the possibility that the rise of artificial intelligence could ultimately make genuine human intelligence, experience and connection more valuable.

An internet worth staying for

Ecologists use the idea of resilience to describe an ecosystem's ability to withstand disruption, adapt and continue functioning. I think that is a useful way to think about where the internet goes from here.

AI isn't going away, nor would I want it to. AI agents will become more capable, synthetic media will become more convincing and the boundary between human and machine activity will become increasingly difficult to see. Trying to return to a pre-AI internet isn't realistic.

But neither is the future of the internet predetermined.

We still decide which communities we participate in, which creators we support, what we publish, what we share and which behaviours we reward. Platforms will make choices about whether they prioritise authentic participation or simply maximise engagement regardless of where that engagement comes from. And those of us building, teaching and talking about AI have a responsibility to think beyond what the technology can do and consider what happens when millions of people start using it at once.

Because perhaps the biggest threat isn't that bots eventually outnumber humans online. They may already do that by some measures, and the internet continues to function.

The more consequential moment would be when humans look around at an internet increasingly populated by machines, stop believing that the people around them are actually people, and decide there isn't much reason to participate anymore.

Which brings me back to the question I started with.

Would you still want to be online if everyone else there was a bot?

I wouldn't.

And that makes keeping the internet human one of the more interesting challenges of the AI era.