Stop free data mining

The Wikimedia Foundation, the non-profit organization that hosts Wikipedia, has called on artificial intelligence companies to stop free-scraping data from the online encyclopedia to train their models. Wikipedia’s request is simple: instead of pushing the website’s free servers, use the company’s paid API, Wikimedia Enterprise.

Wikimedia argues that AI models need high-quality, human-edited information to maintain their quality and efficiency, and Wikipedia, with its extensive network of volunteer editors who keep content up-to-date in more than 300 languages, is one of the most valuable resources in this regard.

Wikipedia’s request from AI companies

The request comes at a time when Wikipedia’s business model is directly threatened by the very technologies that feed off its data. Users are now increasingly asking AI chatbots their questions instead of searching Wikipedia.

On the other hand, Wikipedia (the seventh most visited website in the world) relies on public donations to finance its expenses ($179 million in the 2023-2024 fiscal year) and does not display advertisements. When users use ChatGPT instead of visiting the site directly, they don’t see the donation requests at the top of the page, putting the foundation’s revenue at risk.

Also, the Wikimedia Foundation recently found that unusually high traffic in May and June was caused by AI bots trying to hide their identities and scrape data. This is while human visits to the pages have decreased by 8% compared to last year.

In its statement, Wikimedia argues that Wikipedia’s role as the backbone of knowledge on the Internet is now more important than ever. According to experts, there is concern that artificial intelligence will eventually start to eat itself without first-hand information. Productive artificial intelligence cannot exist without new knowledge created by humans. Without it, AI systems will collapse.

Studies have shown that when AI developers try to remove Wikipedia from their training data, “the resulting answers are significantly less accurate, less varied, and less verifiable.”

Wikipedia’s overall transparency (providing references and verifiable sources and publicly recording all changes) has made it one of the most trusted platforms; Where no algorithm tracks your behavior and everyone sees exactly the same information. In general, Wikimedia urges AI developers to consume its content responsibly. This responsibility includes financial support.

Of course, Wikimedia emphasizes that it is not against artificial intelligence and even uses it to help editors with tedious tasks (such as automatic translation); But these tools should support humans, not replace them.