Wikipedia has urged developers of generative artificial intelligence not to collect data directly from its site, but to use the official paid Wikimedia Enterprise interface. This is reported on the Wikimedia Foundation page.
“Artificial intelligence cannot exist without the human effort that goes into creating open and non-commercial sources of information, such as Wikipedia. That’s why Wikipedia is one of the world’s highest-quality datasets for training AI, and when AI developers try to omit it, the answers they get are significantly less accurate, less diverse, and less verified.
That’s why we urge AI developers and other content users who have access to our content to use it responsibly and support Wikipedia. They can do this through two simple actions: attribution and financial support, ”, the message says.
The Wikimedia Enterprise service provides access to large volumes of content without overloading the servers. In addition, the paid model helps support the foundation’s non-profit mission and ensures that the contributions of authors are correctly reflected in the results created by generative AI.
The foundation emphasized that AI bots have repeatedly tried to covertly collect information from Wikipedia, masquerading as ordinary users. After improving the traffic detection system, it turned out that from May to June a significant part of the requests were sent by AI bots. The report also notes that with a decrease in overall traffic, the number of volunteers editing articles is decreasing, as well as the volume of donations.
Currently, the number of views on Wikipedia has fallen by 8%, as users increasingly turn to artificial intelligence tools. At the same time, Wikipedia founder Jimmy Wales expressed doubts about the ability of AI to create truly encyclopedic texts.
