Editorial illustration for Longtime Wikipedian's "Wake-Up Call" Notes Encyclopedia's Decline Since 2022
Wikipedia's Decline: Longtime Editor Sounds AI Alarm
Longtime Wikipedian's "Wake-Up Call" Notes Encyclopedia's Decline Since 2022
Wikipedia spent years telling itself a reassuring story about generative AI: it would pass through like weather, maybe even leave the place better than it found it. Automate the grunt work, the thinking went, the formatting and citation-chasing, and leave humans to the actual writing. Katherine Maher, who ran the Wikimedia Foundation at the time, warned in a 2019 New York Times op-ed that treating AI as "a scalable replacement for people" was a mistake, not a feature.
Her argument aged well. By 2018, Google was already training systems to churn out "Wikipedia-like articles," and the pressure only built from there. A Princeton study found that roughly 5 percent of nearly 3,000 new English Wikipedia articles created in August 2024 contained substantial AI-generated text.
The tells were almost comic at first: stray em dashes, curly quotation marks, the odd leftover prompt phrase like "Here is your Wikipedia article." Editors tried policing it, fixing what they could, deleting what they couldn't, flagging the fluff that praised some village's "fertile farmlands, livestock, and scenic countryside" alongside sources that didn't exist. What looked like a manageable nuisance turned into something harder to contain.
“The potential ‘whittling down’ each year of our pool of talented writers is the greatest threat to Wikipedia,” another Wikipedian wrote, “as we must increase the number of editors, especially expert editors, to be able to fix the sprawling hellscape of weak, inaccurate and incomplete articles that drag down the project.”
Why this matters
Henner's numbers matter because Wikipedia isn't just another website losing traffic. It's the training data backbone for the same chatbots that may be siphoning its readers. If people get their answers from an AI summary that was built on Wikipedia text, and never click through, the site loses the ad-adjacent attention and donation prompts that keep it funded, while the AI vendor keeps the engagement.
Nobody has to behave badly for this to happen. It's just what happens when a free knowledge commons gets treated as a raw material instead of a destination.
The optimistic 2019-era take, that LLMs would do the grunt work and free up editors for judgment calls, assumed Wikipedia would stay the place people go. That assumption looks shaky at a quarter of lost readership. For anyone building on top of Wikipedia data or citing it in RAG pipelines, this is worth watching closely: the incentive structure that produced the encyclopedia in the first place depends on humans showing up, not just models scraping it. A shrinking volunteer base reading fewer pages is a supply problem for the entire AI stack, not a Wikipedia problem.
Related Reading
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google to End Free Access to Gemini AI for Non-Subscribers
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
From Archives LLMs exceed human agreement on some MT annotation tasks, vary by scheme
From Archives Anthropic Says Claude AI Accidentally Submitted False Murder Tip
Common Questions Answered
What was Katherine Maher's warning about generative AI and Wikipedia in 2019?
Katherine Maher, who ran the Wikimedia Foundation, warned in a 2019 New York Times op-ed that treating AI as "a scalable replacement for people" was a mistake rather than a feature. Her argument proved prescient as Wikipedia has faced significant challenges since the rise of generative AI technology.
Why is the decline in Wikipedia's pool of talented writers considered a major threat?
According to Wikipedians quoted in the article, the "whittling down" of expert editors each year threatens Wikipedia's ability to fix weak, inaccurate, and incomplete articles across the project. The site needs to increase the number of editors, especially expert contributors, to maintain content quality and combat the sprawling problems that drag down the encyclopedia.
How are chatbots trained on Wikipedia content affecting the encyclopedia's sustainability?
When people get answers from AI summaries built on Wikipedia text and never click through to the actual site, Wikipedia loses the ad-adjacent attention and donation prompts that fund its operations while AI vendors retain the user engagement. This creates a problematic cycle where Wikipedia serves as the training data backbone for the same chatbots that are siphoning away its readers and revenue sources.
What specific decline in Wikipedia has occurred since 2022 according to the article?
The article references a "wake-up call" from a longtime Wikipedian noting Wikipedia's decline since 2022, suggesting that the impact of generative AI on the platform has accelerated during this period. This decline appears connected to reduced editor participation and traffic loss to AI-powered alternatives.
Further Reading
- Special report: Wikipedia's declining readership and contributor pipeline - The Signpost
- Generative AI and Wikipedia editing: What we learned in 2025 - Wiki Education
- Generative AI and Wikipedia: The last six months - Wiki Education
- In the age of AI slop and misinformation, we need Wikipedia more than ever - San Francisco Chronicle
- Wikipedia Traffic Drops 8 Percent as AI Replaces Page Visits - Head Topics
