The Convenience Trap

Let’s start with Google’s AI Overviews. These summaries, generated by large language models, now appear at the top of many search results. They distill information from a multitude of websites, offering users instant answers without the need to click through to the original sources. For users, it’s a time-saver. For website owners and publishers, it’s a gut punch.

The traditional web economy was simple: creators published content, Google indexed it, and in return, those creators received traffic—potential customers, ad revenue, and a chance to build their brand. Now, with AI Overviews, that traffic is drying up. Users get what they need from the summary, and the websites that supplied the raw material for the AI’s training are left out in the cold.

The story is even starker in e-commerce. Google’s agentic shopping allows users to find, compare, and buy products directly from the search page. The AI scours the web for the best deals, applies coupons, and completes the purchase—no need to visit the retailer’s site. The merchant loses not just a sale, but also the chance to build a relationship, offer upsells, or gather valuable customer insights.

The Seeds of AI’s Own Destruction

Here’s where the feedback loop begins to look dangerous. If creators and merchants see diminishing returns from their online efforts, what incentive do they have to keep producing high-quality content or maintaining unique product listings? Over time, the web risks becoming a barren landscape of recycled information and commoditized goods.

AI, ironically, is the biggest beneficiary and the most vulnerable victim. These systems are trained on the vast, ever-changing tapestry of the open web. If the web withers—if original content and innovative products dry up—AI will be forced to train on its own regurgitated outputs. This phenomenon, known in technical circles as “model collapse,” means that future AI models could become less accurate, less creative, and less useful.

It’s a bit like a chef who, instead of sourcing fresh ingredients, starts making meals from leftovers. Eventually, the quality suffers, and so does the chef’s reputation.

A Self-Defeating Cycle

This isn’t just a theoretical risk. Early data already shows that websites cited in AI Overviews are seeing fewer clicks, with some reporting drops as high as 60%. Small publishers and independent retailers—often the most creative and innovative players—are especially hard hit. If they can’t survive, the diversity and richness of the web will suffer.

The metaphor that comes to mind is AI “shooting itself.” By prioritizing short-term user convenience and its own ecosystem, AI may be undermining the very sources that fuel its intelligence. Left unchecked, this could lead to a digital ecosystem that’s less vibrant, less trustworthy, and ultimately less valuable for everyone—including the AI itself.

The Road Ahead

What’s the solution? At minimum, there needs to be a re-balancing of incentives. Platforms like Google must find ways to ensure that content creators and merchants are fairly compensated and supported. Otherwise, the web risks entering a downward spiral, with AI models cannibalizing a shrinking pool of original material.

The future of AI—and the open web—depends on a healthy, symbiotic relationship between platforms and creators. If we lose sight of that, we risk building a digital world where convenience comes at the cost of creativity, sustainability, and ultimately, intelligence itself.