AI is now both creator and consumer

Guest Column: Arpit Srivastava, Associate Director with Xiaomi Technology India, writes on how AI content and AI-driven web traffic are reshaping the internet into a machine-to-machine ecosystem

by Arpit Srivastava
Published: Jun 11, 2026 11:33 AM | 4 min read

Cloudflare reports that bots now account for 57.5% of all HTTP requests, surpassing human-generated traffic at 42.5%, marking a historic shift in web interaction dynamics.
An analysis by Graphite indicates that by November 2024, over 50% of new web articles will be generated primarily by AI, a significant increase from just 5% before the launch of ChatGPT.
The rapid rise in bot traffic is attributed to AI agents completing tasks more efficiently than humans, leading to a dramatic increase in web requests and the volume of AI-generated content.
Concerns are raised about the implications of this shift, particularly regarding the potential degradation of content quality and the challenge of maintaining human meaning and significance in an increasingly automated information landscape.

I've been reading Yuval Noah Harari's 21 Lessons for the 21st Century for a few weeks now. He puts forward something that sounds almost sci-fi when you read it: a future where things are created by AI, consumed by AI, and humans gradually become peripheral to that loop. I remember reading it and thinking — yes, directionally this makes sense, but we're probably years away from seeing actual evidence of it.

Then I came across the Cloudflare data. And I had to put the book down.

The number that changes everything

Cloudflare, which serves roughly one in five websites globally, just confirmed that bots now account for 57.5% of all HTTP requests to web content. Human-generated traffic sits at 42.5%. This is the first time in the internet's history that machines have outpaced humans on the web. The Cloudflare CEO himself called it — and he'd predicted it wouldn't happen until 2027. It arrived 18 months early.

At the same time, on the creation side, an SEO research firm called Graphite analysed 65,000 URLs published between 2020 and 2025. Their finding: by November 2024, more than 50% of new web articles were being generated primarily by AI. Before ChatGPT launched, that number was 5%.

So, here's where we are. More than half the content being published on the internet is written by AI. More than half the traffic consuming that content is generated by bots. The loop Harari described is not coming. It's here.

What is actually happening inside this loop

It helps to understand why bot traffic has exploded so fast. When a human shops for a laptop, they might visit five websites. When an AI agent does the same task on your behalf, it might visit five thousand. A single agentic AI completing one task generates a thousand times more web requests than you would. Multiply that across millions of agents running millions of tasks simultaneously, and the maths gets uncomfortable fast.

The content side has its own logic. Publishing AI-generated articles at scale became a cost-efficient SEO strategy. The quality improved rapidly. The volume exploded. The web filled with machine-written text, crawled and indexed by machine processes, to answer queries run by AI assistants. Humans are still somewhere in the chain — asking the original question, reading the final answer — but the middle, where all the production and distribution happens, is increasingly automated.

Where does this go next

The honest answer is: faster than anyone is predicting, based on how the last two years have gone. Cloudflare's CEO predicted 2027. He was off by 18 months. Every forecast about AI adoption speed has been an underestimate.

What worries me more than the speed is the quality of what circulates in this loop. AI models are trained on data from the internet. If the internet is increasingly filled with AI-generated content, those models train on their own outputs. Researchers have a term for this: model collapse. The outputs degrade. The hallucinations compound. The signal-to-noise ratio collapses.

The one thing machines still can't do

Here's what I keep coming back to. AI can produce content and AI can consume it. But AI cannot decide what matters. It cannot feel the weight of a bad decision. It cannot notice that something is missing from a conversation. It cannot bring lived experience to an observation and say — this is significant, this changes something.

That's not a comforting thought designed to reassure professionals that their jobs are safe. It's a more specific claim: humans are still the ones who assign meaning. And in a world drowning in AI-generated content consumed by AI-driven bots, meaning is the scarcest thing.

The question isn't whether humans will be replaced in this loop. The question is whether we'll stay sharp enough to notice when meaning is draining out of it — and say something about it before the loop closes entirely.

Harari asked us to think hard about what makes us irreplaceable. The data is now making that question urgent.

Arpit Srivastava has spent more than 15 years across MNCs and startups working at the intersection of AI, Marketing, Business & Strategy and actively shares opinions from a practitioner's lens

Disclaimer: The views expressed here are solely those of the author and do not in any way represent the views of exchange4media.com.