Wikipedia's struggle in an age of AI answers

A Wired book excerpt looks at declining Wikipedia use as AI summaries answer questions directly, and what that means for the project.

A book excerpt published by Wired examines how Wikipedia is coping with the rise of AI, and argues that the real threat to the encyclopedia may be bigger than lost traffic.

From confidence to concern

The piece describes how the Wikimedia Foundation and Wikipedia's volunteer editors initially expected AI to be manageable, and possibly helpful. Language models might automate tedious housekeeping such as formatting, citations and missing links, freeing people for more creative work. Things turned out to be more complicated.

Early warning signs appeared years ago. Google was already training AI to write Wikipedia-like articles in 2018. A Princeton study in 2024 found that about 5 percent of nearly 3,000 new English Wikipedia articles created in one month contained substantial AI-generated content. Editors learned to spot telltale signs, such as leftover prompt text, and tried to steer AI contributions toward being acceptable or remove them when they were not.

The bigger issue

Now AI summaries in search results and chatbots answer many questions directly, so fewer people visit Wikipedia itself. The excerpt suggests the deeper loss is not page views but the cycle that sustains the project: readers who become editors, and editors who keep knowledge accurate. AI systems rely heavily on Wikipedia's content, yet they can reduce the flow of new volunteers who maintain it.

Why it matters for builders

Anyone building AI answer products depends on healthy sources of human-written knowledge. If AI answers drain traffic and contributors from those sources, the quality of future AI answers suffers too. Good citation practices, with clear, clickable links to original sources, help send readers back to the work that makes answers possible.

For publishers and community sites, the lesson is to give people reasons to visit beyond the basic answer: depth, discussion, tools, original analysis and community. That is also why we write our own context and builder takeaways for each story while crediting and linking the original reporting.

Answer engines and research tools such as Perplexity and NotebookLM are listed in our AI Research & Search category.

Source: Wired. This article is PickAI’s own summary and analysis based on the original reporting.