Who is the internet for?
SEO, Clickless Search, and the AInternet
Imagine designing and building a home while its residents continued living in it. What you create is highly customized to them because you observe them living in real time and make what they need. One day, while you’re still working, these residents move out and new ones move in. Now imagine you didn’t realize that for, say, a year or two afterward.
This is what it has been like to design things for the internet. People lived here once, then AI moved in. But we’re still building a house for people. I think we might be building the wrong thing.
I’ve been designing interfaces for two decades now, and when I look at the modern web, I see a landscape increasingly shaped not by human needs but by machine logic — a vast network of APIs, algorithms, and automated systems talking to each other in languages we never hear. Yes, “we” wrote those languages, but let’s be honest: “we” isn’t most of us.
Last week, my daughter asked me to help her find information about Greek mythology. She’s been reading books about it and had some specific questions that the books couldn’t answer. As we typed in her questions, I noticed something important: Instead of clicking through to websites, we found ourselves staying on the search page as AI-generated answers appeared above the traditional results. Unbeknownst to her, we were witnessing the end of SEO as we know it.
The conventional search engine optimization wisdom is evolving accordingly. The rungs of the SEO ladder aren’t just increasing — making it more difficult to compete for subject matter authority by Page Rank — they’re changing. Specifically, the pattern of SEO to optimizer benefit is going to upend the entire point. I’ve already seen advice suggesting that because Google’s AI synthesizes content differently than a comparatively simple indexing bot, we need to begin to structure our content differently so that it will be more likely to appear in Google’s AI summaries. FAQ content, for a business, for example, could be elevated in this strategy, as its structure anticipates the kinds of questions that people considering a product or service might ask. AI, after all, is already training us to change how we search for things. Specifically, queries are aligning with more conversational semantics rather than metadata-focused keywords and phrases. All fair enough — we can probably gain increased visibility within a search engine’s AI summaries by way of “agentic design.”
But…why?
Old-school SEO had a fairly balanced value proposition: Google was really good at giving people sources for the information they need and benefitted by running advertising on websites. Websites benefitted by getting attention delivered to them by Google. In a “clickless search” scenario, though, the scale tips considerably.
If Google has its way, users will increasingly stay on google.com, their questions answered by AI that synthesizes information from across the web. Yes, there will be an attribution link to your original content, but have you seen them? They are tiny. Who will click them? Our motivation to optimize content for Google is transforming from “please send visitors our way” to “please use our content as a source” — but in this new paradigm, what’s really in it for us? Generally, I’d say… not much.
And if clickless search makes human attention delivery significantly less likely, one has to wonder: will a website’s visual design even matter anymore? How many humans should we expect to actually see what we create? For those of us happy with a very small, human audience, none of this matters much, other than we’ll probably see our numbers continue to drop. For those whose livelihoods depend upon traffic to websites they control and have designed for humans, this matters very much.
So what’s the point of “agentic design”? I can only think of one scenario, and that is when the answer Google’s AI can provide isn’t what you know, but just you. Knowledge about things will go entirely to Google, on our backs. Some knowledge about how to do things a machine cannot will be retained by us.
Perhaps the only content worth optimizing for AI will be that which machines cannot replicate or synthesize — unique human experiences, specialized expertise, creative works that resist automation. Everything else — facts, figures, general knowledge — will be absorbed into the AI’s vast knowledge base, built on our collective work but no longer driving visitors to our individual spaces.
This home we’ve been building is so much bigger than the metaphor can even support. The internet has become a kind of parallel world where machines are the native inhabitants and we humans are more like tourists, guided by AI assistants that translate machine logic into human-readable experiences. Our devices are increasingly less like tools and more like interpreters, mediating our experience of a digital ecosystem that has grown too vast and complex for direct human navigation. And this has all happened very quickly. To be disoriented is understandable.
The interesting question isn’t how to optimize for AI agents, but what kinds of human experiences are worth preserving in a world where machines do most of the talking.
Written by Christopher Butler on
Tagged