Search engines are beginning to move from listing pages toward answering questions with generative AI. If that shift takes hold, people may find information more quickly, but the websites that supply much of the web’s writing could have a harder time attracting readers.
From links to answers
Google is experimenting with AI-powered search after Microsoft brought OpenAI’s technology into Bing. These changes could mark a major shift in how people find information online, after years when search updates often felt incremental.
Large language models, or LLMs, generate answers from patterns learned across huge collections of data. The source article notes that much of the material used to train them is public text from the web. A search engine that can turn that material into a direct response may leave users with less reason to click through to the original pages.
That prospect matters because search traffic helps websites reach audiences. Publishers, bloggers and small software companies may all depend on people finding their work through search results. If an answer appears directly on the search page, those organizations could lose visits and the attention that visits bring.
Publishers face a traffic and revenue problem
Many publications rely on advertising income connected to their audience and web traffic. Fewer visits from search could put pressure on that model. One possible consequence raised in the source is that more digital content could end up behind paywalls as publishers look for other ways to earn revenue.
The concern is not simply that search results may look different. If Google and Bing become more like chatbots than directories of links, search may stop serving as a broad route from readers to websites. That could affect how publications find new audiences and how businesses market their offerings online.
The source also points to a tension in how search companies build these tools: their own services have commercial interests, while independent websites need readers to reach their pages. If search engines provide answers without sending people onward, the interests of the platform and the sites supplying information may not align.
AI-generated writing could complicate the web
Generative AI makes it easier to produce readable text, which could encourage more people to use it for writing. If that output is published publicly and later included in training data, future models may learn from text produced by earlier models as well as from human writing.
That possibility raises questions about the material behind AI answers and how clearly users can tell where it came from. It also creates an unsettling prospect for people who write for a living: systems trained on human-authored work may eventually compete with the writers whose work helped train them.
Publishers may respond by trying to protect their work from being used to train LLMs. If more content is kept away from AI systems or placed behind paywalls, the web could become less open and more divided between services that hold information and sites that try to reach readers directly.
Direct relationships may matter more
If search sends fewer readers to websites, creators may need other ways to keep in touch with their audiences. The source identifies Substack and beehiiv as services built around a cycle of creating writing and distributing it through email lists. A direct connection with readers could become more valuable when search is a less reliable path to discovery.
The larger change is still uncertain, but the stakes are clear. AI search may make it easier to get an immediate answer while making it harder for the people and organizations behind online writing to be found, funded and read. How search companies handle the relationship between generated answers and the websites that inform them will help shape the web’s next chapter.