The jump really does feel unreal. A system that began as a modest way for scientists to share linked documents now answers questions in natural language, breaks them into sub-queries, scans the web on your behalf, and returns a stitched response with citations. That is not marketing varnish on old infrastructure. It is a real shift in how people experience information.
Table of Contents
The web began as a scientific tool with almost no drama
The origin story is almost disarmingly plain. Tim Berners-Lee proposed the World Wide Web at CERN in 1989. By the end of 1990, the first web server and browser were running there, and the first site lived at info.cern.ch. That first page was not trying to entertain, persuade, or hook anyone into a funnel. It explained what the web was, how to create a server, and where to find related resources. In the earliest years, the web was still small enough to feel hand-built.
That matters because the early web was not conceived as a media empire or an advertising machine. It was a practical system for connecting knowledge. The first great miracle of the web was not design, scale, or monetization. It was the simple idea that one document could point to another anywhere else. That alone felt radical.
Then came the decision that changed everything. On 30 April 1993, CERN released the web to the public. It did not treat it as a closed product or a licensed network. It opened it. The web became enormous because it was useful, open, and easy to adopt before anyone had fully learned how valuable it would become.
That is the world people are usually picturing when they talk about “1 KB pages.” The exact number varied, but the instinct is right. Pages were tiny, mostly textual, structurally simple, and startlingly light by modern standards. They loaded fast not because engineers had perfected performance, but because there was barely anything there. A page was often closer to a digital note than a full software environment.
Browsers gave the web its first real magic trick
The first big leap was visual. NCSA Mosaic became legendary because it made the web far easier to use and because it was the first widely published browser to display images together with text in a seamless way. That sounds ordinary now. In 1993, it changed the emotional texture of the medium.
A linked document system started to feel like a place people might actually want to spend time in. The web stopped looking like a specialist’s tool and started looking like a public medium.
Soon after, the stack thickened. CSS made it possible to separate presentation from structure in a far more deliberate way. JavaScript gave pages behavior. The web stopped being only a library of documents and started becoming an application platform. Layout, interactivity, forms, state, dynamic updates, and eventually browser-based software all grew from that turn.
Once that shift took hold, the internet was no longer just a place to publish pages. It became a place to build products, services, identities, businesses, communities, and economies. That is where the old web starts to disappear in the public imagination. The bones are still there. The experience changed so much that it feels like a different invention.
Search turned chaos into something people could use
As the number of pages exploded, the next problem became obvious. A web without good search becomes a junk drawer at planetary scale. Early browsing worked when the network was small enough to explore through directories, recommendations, and curiosity. That stopped scaling very quickly.
Google’s PageRank era changed the logic of discovery. A link was no longer just a path from one page to another. It became a signal. The structure of the web itself started to help determine what was worth seeing first. Search was no longer just matching terms on a page. It was beginning to infer authority, relevance, and relationship.
That shift is easy to underrate because people got used to it so fast. Typing a fragment of thought into a search box and receiving something useful in a second came to feel normal. It was not normal. It was one of the defining technical achievements of the modern internet. For millions of people, search became the real front page of the web.
Pages grew from kilobytes into software payloads
Here is where the science-fiction feeling sharpens. While discovery became smoother, the pages themselves became far heavier and more complex. HTTP Archive’s 2025 Web Almanac shows that the median mobile home page grew from 845 KB in July 2015 to 2362 KB in July 2025, while the broader trend in its dataset runs from 505 KB on mobile in October 2014 to 2559 KB in July 2025.
That is not a small increase. It is a change in what a web page is expected to carry.
A modern page often arrives with large images, video, custom fonts, JavaScript frameworks, tracking scripts, recommendation widgets, third-party embeds, interface libraries, and personalization layers. The old web asked a browser to fetch a document. The current web often asks it to boot a miniature operating environment.
That is why nostalgia for tiny pages still lands so hard. The comparison is not sentimental fluff. It points to a genuine transformation. Early pages were lightweight because the web had not yet learned to simulate so much. Today’s pages are heavier because the browser became the universal container for media, software, commerce, publishing, and interaction.
A small comparison that explains a huge shift
| Layer | Early web | Current AI-search web |
|---|---|---|
| Core experience | Read linked documents | Ask for synthesized answers |
| Discovery model | Browsing and simple keywords | Ranking, retrieval, reasoning, follow-up prompts |
| Typical page weight | Tiny and mostly text-based | Often media-heavy and script-heavy |
| User expectation | Find a page | Get a result, then verify it |
This is the heart of the transformation. The web did not just get bigger. It developed layers of abstraction. First came documents. Then visual interfaces. Then applications. Then ranking systems. Now answer engines sit above the page and compress the open web into a more conversational experience.
The web no longer waits for you to click
AI search changes the unit of value. Traditional search mostly sent you toward documents. AI search often tries to assemble the first usable version of the answer before you ever open a page.
Google says AI Overviews are available in more than 200 countries and territories and in more than 40 languages. It has also described AI Mode as a more advanced search experience that uses reasoning, multimodal understanding, follow-up interaction, and a query fan-out approach that breaks a question into subtopics and launches multiple searches in parallel.
OpenAI’s search experience moves in a similar direction through conversation. Microsoft’s Copilot Search in Bing does the same from another angle, blending conventional search with AI-generated synthesis and cited links. Different companies are converging on the same assumption about user behavior: people increasingly want interpretation first and navigation second.
That is a profound change. The search box used to be a doorway. It is becoming a respondent.
AI search still depends on the open web beneath it
There is a temptation to describe this moment as if pages no longer matter. That would be a mistake. AI search does not replace the web. It compresses the web into an answer layer. The sources still matter. Reporting still matters. Documentation still matters. Original writing still matters. The answer has to come from somewhere.
This creates a strange dual reality. On the surface, users see less of the old ritual of browsing from page to page. Underneath, the system still depends on the open web to supply facts, evidence, perspectives, and primary material. In one sense, the page has been demoted. In another, it has become more foundational than ever because it is now raw material for machines as well as humans.
That tension is likely to define the next phase of the internet. Publishers want visibility. Users want convenience. Platforms want trust, scale, and retention. AI search sits right at the collision point.
The smoother the answer looks, the more careful people may need to be
This is the part where awe needs company. Google’s own documentation notes that AI Mode does not always get it right. That warning is not some legal afterthought buried in the margins. It belongs at the center of the conversation.
The new interface is smoother, more fluent, and more complete-seeming than the old one. That can make mistakes harder to detect. Traditional search often forced a little skepticism because the user had to compare sources manually. AI search can make uncertainty look tidy. The danger is not only error. It is error wrapped in confidence and convenience.
That does not make the shift less impressive. It makes digital literacy more demanding. The next generation of internet skill may not be about learning how to search. It may be about learning when an answer is good enough, when it needs verification, and when the source beneath it matters more than the summary on top.
The scale of the internet makes the whole story even wilder
The sci-fi feeling is earned partly because the infrastructure now sits at a civilizational scale. According to the ITU’s 2025 figures, almost three-quarters of the world’s population are online, while 2.2 billion people remain offline and many more face serious gaps in affordability, speed, or quality of access.
So this is not merely a story about nicer interfaces. It is a story about a global information layer that shapes education, commerce, politics, culture, entertainment, research, and everyday decision-making for billions of people. The web is no longer a technical curiosity. It is part of the operating texture of modern life.
That makes the arc from tiny pages to AI answers even more striking. We are not talking about a niche tool that became more powerful. We are talking about one of the core systems through which humanity now finds, shares, filters, and interprets knowledge.
The science-fiction feeling is not exaggeration
People are right to feel stunned by the distance traveled. The web began with sparse text, minimal styling, and a belief that linking knowledge might be enough. Then it learned to display images, run code, rank the world, carry full applications, and respond in natural language while searching behind the curtain.
Seen step by step, the progression looks rational. Seen all at once, it looks impossible.
That is why the move from kilobyte pages to AI search feels like science fiction. Not because it happened suddenly, but because each stage quietly redefined what the internet was for. We did not merely build bigger websites. We built a system that moved from publishing information to organizing it, then from organizing it to interpreting it back to us.
And the strangest part is that the old web is still there under everything. Pages. Links. Servers. Sources. Human beings writing things down and hoping to be found. The machinery above it has become astonishing. The foundation still looks familiar.
Author:
Jan Bielik
CEO & Founder of Webiano Digital & Marketing Agency
This article is an original analysis supported by the sources cited below
The birth of the Web
CERN overview of how Tim Berners-Lee invented the World Wide Web in 1989 and how the project began.
https://home.cern/science/computing/birth-web
A short history of the Web
CERN timeline covering the first server, first website, early browsing, and the web’s earliest public development.
https://home.cern/science/computing/birth-web/short-history-web
30 years of a free and open Web
CERN account of the public release of the web on 30 April 1993 and the importance of making it open.
https://home.cern/news/news/computing/30-years-free-and-open-web
NCSA Mosaic
University of Illinois NCSA background on Mosaic and its role in making the web more visual and accessible.
https://www.ncsa.illinois.edu/research/project-highlights/ncsa-mosaic/
CSS history
W3C historical summary of CSS beginning at CERN in 1994 and becoming a W3C Recommendation in 1996.
https://www.w3.org/Style/CSS/history.html
JavaScript
MDN reference describing JavaScript as the scripting language most closely associated with web pages and browser interactivity.
https://developer.mozilla.org/en-US/docs/Web/JavaScript
The Google PageRank Algorithm
Stanford material explaining the PageRank concept that helped search evaluate the importance of web pages.
https://web.stanford.edu/class/cs54n/handouts/24-GooglePageRankAlgorithm.pdf
A guide to Google Search ranking systems
Google Search Central documentation on modern ranking systems and how Google evaluates indexed content.
https://developers.google.com/search/docs/appearance/ranking-systems-guide
Page Weight
HTTP Archive Web Almanac chapter documenting the growth of page size across the modern web.
https://almanac.httparchive.org/en/2025/page-weight
AI Overviews expand to over 200 countries and territories, more than 40 languages
Google announcement on the scale and rollout of AI Overviews in Search.
https://blog.google/products-and-platforms/products/search/ai-overview-expansion-may-2025-update/
Google Search AI Mode update
Google explanation of AI Mode, query fan-out, multimodality, and follow-up search behavior.
https://blog.google/products-and-platforms/products/search/google-search-ai-mode-update/
Get AI-powered responses with AI Mode in Google Search
Google Help documentation outlining AI Mode’s capabilities and stated limitations.
https://support.google.com/websearch/answer/16011537
Introducing ChatGPT search
OpenAI announcement describing conversational web search with timely answers and links to relevant sources.
https://openai.com/index/introducing-chatgpt-search/
ChatGPT search
OpenAI Help Center documentation on ChatGPT search availability and usage.
https://help.openai.com/en/articles/9237897-chatgpt-search
Introducing Copilot Search in Bing
Microsoft Bing announcement describing Copilot Search as a blend of traditional search and generative AI with cited links.
https://blogs.bing.com/search/April-2025/Introducing-Copilot-Search-in-Bing
Facts and Figures 2025
International Telecommunication Union report on global internet adoption, remaining offline populations, and connectivity gaps.
https://www.itu.int/itu-d/reports/statistics/facts-figures-2025/
