At 25, Wikipedia faces a double threat: the rise of AI and the decline of local media

When Wikipedia number one emerged in 2001, it was allay a clip when to the highest degree had to be patient for info — waiting for the high-pitched scree and its answering cry as the computer connected, painstakingly, to the internet via dial-up. 

And the idea of an open source encyclopedia that could be updated by anyone in real time — or its equivalent in those pre-fibre-optic days — sparked questions and plenty of criticism about how accurate that information could be. 

Fast-forward 25 years and Wikipedia is now the ninth most visited site on the internet, with nearly 15 billion visitors each month, searching and editing its more than 65 million articles.   

But despite its speedy ascent in the early years and steady growth thereafter, Wikipedia isn’t as visible as it used to be. Now, when you Google a question, the top search result will likely be a Wiki link, but its AI will also handily synthesize the answer for you above it. And ChatGPT? That cuts Wikipedia out altogether.

Now, human visitors to the site are on the decline, dropping by roughly eight per cent in parts of 2025, while large language models (LLMs) — chatbots or other forms of AI that can condense words and information — are hammering Wikipedia’s servers and using it as a training ground. 

If these trends continue, alongside the decline in local news outlets that are Wikipedia’s main sources, the future is “more dire than you think,” says Zachary McDowell, an associate professor of communication studies at the University of Illinois in Chicago and the author of Wikipedia and the Representation of Reality.

Look at it like a pyramid of information accessibility, with LLMs at the top, Wikipedia in the middle and traditional news media on the bottom, he said.

“As you erode all the secondary sources below and then you start to erode Wikipedia, what you have is something that will inevitably crash in upon itself,” he said.

“It’s been shown over and over again that when you feed these [AI] systems synthetic data, when you feed them things that have been created by other AI sources, they end up with what they refer to as model collapse.” 

In layman’s terms, it’s considered digital inbreeding — when AI-generated information gets fed back in on itself again and again, increasing the number of errors and inaccuracies.  

Wikipedia’s founder, Jimmy Wales, expresses more concern about the financial implications of the increasing demand that LLMs are placing on the online encyclopedia. He notes the need for more databases and servers to support that extra traffic from “AI crawlers” was the reason behind the deals it announced with several AI partners in January, including with Amazon, Meta and Microsoft. 

“The average donation to Wikipedia is about $10 [US],” he said. “People aren't donating to subsidize OpenAI."

But McDowell’s concerns about those AI crawlers making it more difficult to access neutral, accurate information? Wales said he doesn’t share them when it comes to Wikipedia. 

“We don’t listen to AI; Wikipedia is written by humans and one of our strongest policy points is that everything in Wikipedia needs to ... Have a quality source,” he said. “That’s the pathway into Wikipedia … human-created, human-vetted knowledge.” 

But McDowell and Wales agree that media concentration — especially in small local newspapers and news stations — affects Wikipedia, but in a larger sense, it also affects the ability to accurately capture a record for history.   

Conglomeration erodes the “neutrality” for which Wikipedia and traditional media strive, McDowell said. 

“These conglomerates, many of which have very political leanings, are now pushing a particular ideology and agenda."

Can you spot the deepfake? How AI is threatening elections

In Canada, more than 250 local news publications or broadcasts have shuttered between 2008 and Oct. 1, 2025, according to the advocacy group News Media Canada. 

“That first draft of history isn’t even being captured in the first place. So there's no question that that's a problem.” 

AI, however, is just speeding up what McDowell calls “the Wikipedia detour” — something that began a decade ago, as Google started summarizing answers on the search results page itself.  

Cutting Wikipedia out of the equation doesn’t just affect its ability to recruit editors or donors, it undermines digital and information literacy, because people don’t see the citations that form the foundation of these articles.  

Nor are they encouraged to dig deeper, in the way that what might start as a search about black holes eventually brings you to the dates for an upcoming lunar eclipse. Wikipedia can be a rabbit hole, but in a good way.

It’s how Jess Wade has helped boost the profile of female scientists. The British physicist and assistant professor at Imperial College London in the U.K. Has written more than 2,200 biographies on Wikipedia of women and other marginalized groups who work in the sciences over the past eight years, saying that most of her articles get visited as people are investigating a scientific concept and then stumble upon the fact that it was invented by a woman. 

And that boosts their visibility in real time. During the COVID-19 pandemic, Wade and a colleague added biographies about the women and people of colour on the front lines of the public health crisis. As time wore on, she said, the “old white men” who had been appearing in newspapers or on TV began to be replaced by some of the experts she had included.

“I was really struck by how many broadcasters or teachers or lawyers use Wikipedia as a first point of call when looking for information."

There are ways, though, that Wikipedia is exploring how to use AI to improve, including its search experience, as its interface hasn't changed much in recent years. That could include using a chatbot, Wales said.

And while the site’s 250,000 volunteer editors would still be the ones curating it into the future, he said he can see AI doing some simple automation — fixing a dead link in an article, for example, by finding potential replacements that a human could validate and decide whether to include.

“Automating some of the drudgery of working on Wikipedia could be very helpful and sort of make it higher quality.”

Senior writer

Radio segment produced by Andrea Hoang

---

---