AI Search Could Kill the Web That Feeds It

I'm going to be up front and admit that I genuinely don't know what will happen, but I've been spending more time thinking about this ever since Google started pushing AI search.

AI search results fundamentally change the economic agreement that created the internet we have used for the last few decades. And by fundamentally altering that agreement, I think that Google expects its customers and users to act irrationally.

If you have predictions, feel free to leave a comment below.

The Old Google Relationship

Most people understand how search works from a user perspective:

• Google crawls and indexes most of the public internet, preparing the data for search.
• A person enters a search query.
• Google returns a set of ranked choices.
  • These rankings are imperfect because ads exist, Search Engine Optimization (SEO) plays a role, and Google has some proprietary things going on. Regardless, a ranking is returned.
• The user gets to compare sources and decide who to trust and which sites to visit.
• The site owners get the chance to earn traffic, reputation, leads, sales, and ad revenue.

Ultimately, this means that human attention is the goal and what everyone is optimizing for. AI overviews strip this away.

The AI Summary "Solution"

An AI summary isn't really a search result. Where before, Google was an indexing and ranking service, AI summaries make Google an answer service. Sure, it tries to cite the sources, but citations don't generate clicks. In fact, Ahrefs found that the presence of an AI overview makes click-through rates drop by over 50%.

This is radically different than the old search. And, more importantly, a footnote isn't a business value generator.

And as someone who creates content and provides education for a living, I have a question that I haven't seen a good answer to...

Who Pays for Source Material?

Trust me when I say that original content is expensive. It requires research, expertise, writing, fact-checking, copy editing, and hosting, among other things. It's always interesting when I quote companies on custom learning content, because they're frequently surprised at how much truly original, tailored content costs. Because it takes a team to do well.

Many free information sites survive on ad revenue. When they get lots of visitors, they show ads, and those ads reward them for clicks and impressions.

But with AI summaries, the AI can answer the query without sending a visitor to the creator. Google seems to think this is fine, because they're Google, and I don't think they can imagine a world where 90+% of all search traffic doesn't go through them. Most regular people don't realize how much of a stranglehold Google has on the market. If they cut off your business from search, it's likely dead.

So sites are willing to put up with the ad buys, playing the SEO game, and other issues that come from one company dominating search. But Google appears to be betting that creators will behave irrationally. Specifically:

• Creators will continue publishing anyway
• Users will trust opaque summaries over source evaluation
• Businesses will keep paying for visibility in a system designed to reduce exploration and discovery
• The public web will remain full of original human material

I think they're wrong. Very wrong.

The Dead Internet Problem

Dead Internet Theory started as a conspiracy theory claiming that the web is increasingly dominated by bots and manipulated content. I am not endorsing all of that theory. But some of the trends are real: bot farms manipulate political sentiment, coordinated campaigns promote or attack products, and generated content is flooding the public web.

Recently, Cloudflare's CEO said that AI bot traffic will exceed human traffic in 2027, and I think he's right.

If this continues, we'll end up in a situation where creators will have three choices:

1. They pay the high expenses for original content, investing in high-quality content for minimal gain (unlikely anyone does this in the long term).
2. They start producing slop and try to out-slop/out-SEO others (many are already doing this).
3. They just stop producing public content and find different avenues to get human traffic (this is what I think will happen).

In all three scenarios, Google destroys what makes them valuable. Additionally, if the original creators leave or paywall, the models and bots will increasingly have to train and retrieve from synthetic output. How well is that going today? Not well, as Anthropic is currently buying old books to scan because when AI trains on AI output, it gets dumber.

So, not only is Google actively harming the companies that make its search product useful, it's actively degrading the pool of original material that the AI industry needs to train its models in the future.

Search is Valuable Because of Discovery

Sure, traditional search was about finding answers, but I believe surfacing discovery was nearly as important as the answers themselves. When you scroll through search results, you have the opportunity to find unexpected connections and information sources.

I can't tell you, over the course of my life, how many products, services, and sites I've found "by chance" on search. There are so many small businesses, experts, products, publications, and communities out there, and it was an awesome feeling anytime I found one of these gems.

An AI answer engine takes that discovery away. And no, citations don't count because, as I linked above, people don't click on them. What it does is narrow things down to the "most likely" response. Efficient? Sure. Good for business? Only if you have the money, authority, or luck to be included in the AI summary.

And that's the rub: these systems are optimized to surface what they already recognize as relevant, authoritative, and statistically likely. So how the fuck is a new business supposed to get noticed? Because a site used to have a shot at being found because people could scroll through results, click a new link, and decide for themselves if the content was useful. Now, it will be preselected by an opaque system before a human ever sees it. It can and will be abused.

So when I see Time Magazine creating content specifically for bots, I see the death of the human internet. I see companies creating massive amounts of slop to try to stay relevant in AI summaries. I see them using bot farms to push narratives. Even more fake reviews, more astroturf blogs and listicles. This has already been a thing for a while. AI is going to make it much worse.

What's Next?

I really don't know. What I do know is that few people will be willing to create and post content if there's no incentive to do so. And I know that Google and Meta cannot survive without human eyes, because even if bots take over the internet, humans decide where to spend their money.

Tread carefully, hyperscalers.