The Robot Read Your Article So You Don't Have To (And Now Nobody Will)

There's a particular kind of irony baked into the modern search experience. You type a question into Google. A glowing box at the top of the page — friendly, confident, suspiciously well-formatted — answers your question in three sentences. You close the tab. You have received information. No journalist got paid. No website got a visit. No link was clicked. The system worked perfectly, depending entirely on your definition of the word worked.

Welcome to the AI summary era, where the web's original social contract — you make content, people come to see it, everyone benefits — has been quietly shredded and replaced with something that benefits roughly two companies and a handful of venture capitalists.

The Middle-Man Became the Destination

For the better part of three decades, search engines operated on a relatively honest premise. They indexed the web, ranked the results, and pointed people toward the actual sources. It wasn't perfect. Google's results page slowly transformed into something resembling a Times Square billboard arrangement, with ads and featured snippets crowding out organic links. But at least the underlying idea — go to the place that has the thing — was still nominally intact.

Then came AI Overviews. Then came ChatGPT's browsing mode. Then came Perplexity, which built an entire business model around summarizing other people's reporting and presenting it with the confidence of someone who definitely did their own research.

The result is a layer of algorithmic middlemen who have positioned themselves between every curious person and every piece of original content on the internet. They read the article. They digest the article. They present you with a smoothie version of the article. The original article sits there, unvisited, slowly composting.

Publishers Are Running a Marathon with No Finish Line

Here's where it gets genuinely grim for anyone who cares about, say, knowing things that are true. Journalism — real, reported, someone-actually-made-a-phone-call journalism — costs money to produce. That money has historically come from advertising, which is tied to traffic, which is tied to people actually visiting websites.

AI summaries are systematically severing that relationship.

Several independent publishers have reported traffic drops of 20 to 40 percent since Google rolled out AI Overviews at scale. The Wall Street Journal reported in 2024 that major news sites were seeing significant referral traffic losses from Google. Smaller outlets — the ones covering local government, regional business, niche topics that nobody else touches — are getting hit hardest because they have no brand recognition to fall back on. Nobody is typing smalltowngazette.com directly into their browser. They were surviving on search traffic. Past tense.

The cruel twist is that AI systems were trained on that same journalism. The summaries sound authoritative because they're distilled from years of reported work. The machine ate the meal and now it's describing the taste to everyone who walks by the restaurant.

The Headline Arms Race Nobody Wanted

So what do publishers do? They adapt. And the adaptation is making everything worse.

If a straightforward, accurately-headlined article gets summarized and bypassed, the incentive shifts toward producing content that cannot be easily summarized — or at least content so provocative, so emotionally loaded, so aggressively specific that a reader feels compelled to click through anyway. This is how you get headlines that read like they were written by someone who just drank three espressos and watched cable news for six hours straight.

The headline arms race was already happening before AI summaries entered the picture. But the new dynamic has poured gasoline on it. If the robot is going to intercept your traffic, you have to make the headline itself the hook — the emotional payload has to land before anyone clicks. Which means more outrage. More hyperbole. More manufactured stakes. More "you won't believe" energy applied to stories about municipal zoning decisions.

This is not a coincidence. It is a direct behavioral response to an incentive structure that tech companies built and then handed to publishers like a gift with a hidden bill attached.

The Discovery Engine Is Broken

There's a subtler loss here that doesn't show up in traffic analytics, and it might actually be the more significant one.

The early web was a discovery machine. You searched for one thing and found three other things you didn't know you needed. You followed links. You stumbled into corners of the internet you never would have found otherwise. The hyperlink was a genuine act of generosity — someone saying here, this other place is worth your time.

AI summaries are structurally anti-discovery. They are designed to terminate the query. The goal is resolution, not exploration. When the AI answers your question in a box at the top of the page, the implicit message is: you're done here, move along. There are no rabbit holes to fall into. There are no unexpected connections. There is no serendipity. There is only the efficient delivery of a pre-chewed answer and the quiet death of curiosity as a navigational tool.

The web used to be a place you explored. Now it's increasingly a vending machine. You put in your query, you get your answer, and the transaction is complete. The fact that the answer was assembled from sources that will now go unfunded and eventually stop producing new information is, apparently, someone else's problem.

Who Actually Wins Here

Google wins. Microsoft wins. Anthropic, OpenAI, and Perplexity win — at least in the short term, before the content ecosystem they're strip-mining collapses and they have nothing left to summarize.

Users get a slightly more convenient search experience right up until the point where there's nothing left to search for because nobody can afford to report it anymore.

And publishers? Publishers get to watch their traffic graphs trend downward while being told by the same companies harvesting their work that they should feel grateful for the "exposure."

The exposure, as it turns out, is the kind you get from standing outside in January without a coat. You're very visible. You are also freezing to death.

A Web That Eats Itself

The internet has always had a self-destructive streak. Platforms that needed user content to survive eventually made life miserable for users. Comment sections that needed community to thrive eventually burned down the community. This is basically the BadrWeb beat — watching the web find increasingly creative ways to undermine the things that made it valuable.

AI summaries are just the latest and perhaps most elegant expression of that tendency. The technology is impressive. The summaries are often accurate. The convenience is real. And the long-term consequence is a web that progressively hollows itself out, optimizing for the frictionless delivery of information right up until the moment there's no information left to deliver.

At some point, the robots will have nothing left to summarize. They'll look at each other. Then they'll summarize that.