Over the past few decades the Internet has gone through several transformations: from a niche space for tech/scientific users, to a mainstream social hub, and now increasingly toward a machine-operated environment.
Each phase has brought both progress and trade-offs. And with today’s AI advancements, we’re entering a new stage that may fundamentally change what the internet is and who it’s for.
With the growing adoption of AI-generated content, search engine answers and AI agents that consume that info, the need to visit websites is becoming unnecessary, we’re seeing early signs of what some call the “Dead Internet” theory: a web where content is largely produced, managed, and consumed by machines and not humans.
This, to me, is the real Web 3.0 (not the version hyped through crypto, tokens, NFTs and blockchain) but a new generation of the web dominated by generative models, algorithmic curation, and minimal human authorship. Which worries me, not in a dystopian or alarmist sense, but in a practical, long-term one: What happens to the open web when it no longer needs people to contribute? Or worse, makes it no longer worth the time and effort for people to contribute their knowledge? You don’t need to go very far, this blog you’re at right now may one day be just my personal diary.
The early Internet: A web built by experts
The early internet was created and maintained by people with technical knowledge and purpose. Publishing content in the 90s or early 2000s required real effort: hand-coding HTML, running your own server, understanding domain registration and writing with clarity for a mostly technical audience (the first website I made was hand-coded in the initial versions of HTML, the Windows 3.1 Notepad.exe and weighed just a few KBs).
That barrier to entry shaped the kind of content and quality we saw online. Forums like Usenet, early blogs, and personal websites often reflected deep knowledge, specialized interest, or a desire to share insights with others. It wasn’t about monetization but about curiosity, contribution, and community. I sometimes come up with articles I wrote in this blog nearly 20 years ago and man, have things changed!
Search engines like Google played an important role by surfacing this knowledge to a wider audience. In return, websites were rewarded with visibility and sometimes modest income through ads or affiliate links. The ecosystem, while imperfect, was relatively balanced.
The social media shift
By the late 2000s and into the 2010s, platforms like Facebook, Twitter, and Instagram changed everything by moving the focus from information to engagement. Content creation became drastically easier, but also less intentional: rather than building long-form resources or publishing on personal websites, users began posting short updates optimized for likes, shares, and algorithmic reach.
This lowered barrier was both a democratizing force and a double-edged sword. On the one hand, it allowed billions to participate in online conversations. On the other, it flooded the web with noise (repetition, outrage, misinformation and distraction). Much of the thoughtful, evergreen content of the early web was buried under a sea of disposable media.
Still, the web was alive. It was human-driven. Content came from people, even if the quality varied widely, you could still curate your preferred experience.
Enter the machine-operated web
Today, a new shift is happening at hyperspeed. In just a few months compared to what took years or decades, the rise of large language models (LLMs) and AI tools integrated into search engines (such as Google’s AI Overviews) are changing how people access information. Users are not visit websites any more. Most queries nowadays are simply asking a question and receiving an AI-generated summary “stolen” from hundreds or thousands of sources behind the scenes.
At first glance, this seems efficient and a great advancement, but it has significant ripple effects.
- Website traffic declines because users get answers without clicking
- Content creators receive no attribution, feedback, or compensation
- New content becomes harder to sustain, especially if it’s time-intensive or costly to produce
- The incentive to publish openly on the web diminishes
If fewer people are writing, discussing, or sharing new knowledge online, we may eventually reach a point where the visible internet becomes a kind of stagnant archive: useful, but no longer evolving. At the same time, AI systems trained on existing content will keep generating new answers, even if the underlying knowledge base stops growing, and simply generating more noise and repetition that doesn’t add any value to humanity.
That’s the essence of what some refer to as the “Dead Internet” idea. Not that the internet disappears, but that its vitality and originality are replaced by automation and passive consumption.
As a personal example, this blog you are reading right now has lost 75% of its visitors in the last couple years:
Google’s role in the transition
Search engines (especially Google) are accelerating this shift. For many years, Google connected users to web pages. But as of 2024-2025, likely motivated by the pressure to maintain user engagement and shareholder value, its interface has started answering questions directly using AI, often summarizing from multiple sources without requiring a single click (and they are somewhat worried about this, but remember that Google is an Ad company, not a Search Engine, so this doesn’t really affect them much as long as they can keep serving ads elsewhere).
The benefits for users are clear: faster answers without the need to surf multiple sites and often a smoother and more consistent experience. But the downside is structural: content producers lose visibility and traffic, even as their material is used to power those answers.
This change subtly transforms the relationship between the web and its most influential gatekeeper. Google and all other LLMs no longer simply index the web but compete with it. And in doing so they could unintentionally erode the very content ecosystem they depend on.
What’s at stake is not just traffic, but purpose
Even if we forget about SEO or ad revenue, the broader role of humans in the digital information landscape changes drasticlly.
- Will people still write thoughtful blog posts if no one reads them?
- Will communities still answer questions in forums if the answers are quietly extracted and used elsewhere?
- Will small publishers maintain open websites if their content is never credited or visited?
When the primary interface to the internet is a machine that paraphrases from other machines, the open web becomes less relevant. And when the incentives to participate vanish, the open web risks fading altogether.
That’s not to say that AI should be avoided. It’s a powerful tool that I use on a daily basis, and when used responsibly, it can enhance understanding and access. But if it becomes the default tool for online knowledge we risk turning the web from a living commons into a read-only utility.
How do we preserve human participation?
The challenge is not to stop progress but to balance automation with sustainability. If AI and search engines become the dominant way people interact with information, then they must also help sustain the system that feeds them.
- Attribution and linking: AI-generated answers should include clear links to source content
- Compensation models: Search engines and AI companies should pay licensing fees for content they use (even pennies on the dollar would go a long way)
- Interface nudges: Tools could guide users to explore full articles or discussions rather than just accepting summaries
- New formats: Websites may need to evolve toward more interactive, real-time, or community-driven experiences that AI can’t replicate easily
Above all, we need to ensure that humans remain involved not just as consumers but as creators and contributors. The open web only works if people still care enough to share their knowledge freely.
A quiet turning point
The internet isn’t fully dead but it is changing into something quieter, less participatory, and more machine-managed than it once was.
Whether that’s a good or bad thing depends on how we view it and respond. If we value the diversity, depth, and serendipity that came from people publishing for other people, we’ll need to find ways to protect and incentivize that activity, even in an AI-saturated future.
If we don’t, we may end up relying on machines to approximate what humans once shared directly, losing original content, creativity and the warmth of online communities along the way.
No comments yet