Wikipedia is exploring the roles artificial intelligence (AI) systems will play in the future of the volunteer-run online encyclopedia, the site’s founder Jimmy Wales said in a recent interview with The Hindu.
“We’re having a lot of conversations about AI at the moment, so that’s a lot of the intellectual work we’re doing these days, thinking and talking to people in the community about what that means for the future,” Mr Wales said.
Wikipedia’s traffic, at least from human readers, has fallen by about 8% over the past few months. The decline was due to artificial intelligence models, such as those built into Google Search, which repackage accurate user queries with answers gleaned from Wikipedia and other sources. Because the site is ad-free, a drop in human traffic doesn’t affect it like it would a news site, Mr. Wales said, adding that the Wikimedia Foundation has taken steps to share articles with AI developers in a responsible manner.
Mr Wales said that even before large language models (LLMs) were introduced, “Google started to get smarter and smarter and then it could tell you how old Tom Cruise is” without users having to visit a Wikipedia page. Now with LLM, “quick questions” are answered without driving traffic to Wikipedia.
He stated that if the source of the information remains unknown to the user, it may affect donations to the site. “For us, that makes a difference,” he said. “Obviously, if people don’t know the information is from Wikipedia, they don’t remember, ‘oh, I’d like to contribute to support Wikipedia,’ that would be a problem. So attribution is really important. Accurate page views aren’t really something we obsess over.”
AI Licensing
Last year, Wikimedia Enterprise published a dataset of its English and French articles on the AI-friendly Kaggle platform so that LLM developers wouldn’t just download data from its site. Some volunteers expressed concern about their contributions being part of the LLM training data. Mr Wales said: “I’m sure there are some (complaints) because it’s obviously a big, noisy community.”
Mr Wales pointed out that Wikipedia’s content was provided under a Creative Commons licence, which had always allowed for both non-profit and for-profit commercialisation. “CC-BY-SA (a license mandating attribution and a ‘share-alike’ requirement mandating that those who modify it apply the same license) is like open source software,” he said. “So you are free to copy, modify and redistribute versions… We have not granted any new rights to AI to use this data; they always had that right.”
Indian languages and quality translation
He said the Indian language Wikipedia community could see some opportunities from AI translations. However, this could be somewhat complex and require a lot of human supervision, he added.
Recalling the poor quality of translations in the past, Mr Wales said that “the objection (to machine-translated Wikipedia articles) was not some kind of hatred of machines or technology…” but of quality. “It was like, wow, the translations are really bad, so you shouldn’t do that.”
An “experiment” with translations into a small Austronesian language—probably Cebuano, which uncharacteristically rivals the major world languages in terms of articles—resulted in substandard articles in that language.
Pointing to experiences across other languages, he said poor content is not an ideal platform to build a community on. “It’s almost like an SEO (search engine optimization) strategy, right? People would click and they’d see it and they’d see bad content, but they could edit it and make it better. So would that be a way to start a community? It turns out the answer is no. It didn’t grow the community. And actually it was probably bad because in that language, people were probably like, yeah, Wikipedia.”
Mr Wales said it would be a welcome development if AI “could increase the productivity” of contributors across languages. However, “even good machine translation can easily (fall short of) traditional standards” if “cultural issues” are not addressed, he said.
Legal battles and censorship
Wikipedia is fighting court cases around the world and has faced off against news agency ANI in India. It took down an English Wikipedia article globally for the first time, breaking the encyclopedia’s traditional resistance to censorship. But Mr Wales suggested it was done as part of a tactic to wage a legal battle.
“Our view is that the fundamental right of access to knowledge is really, really important to us, and to have something truthful and neutral,” Mr Wales said. “Obviously what we have to do when we have a legal situation is that you have to maneuver and you have to be very careful not to do anything that eliminates your right to proceed to the next stages.”
Published – March 7, 2026 07:15 IST
