Wikimedia makes its AI’s friendly data

Wikimedia, non -profit behind Wikipedia sites and sister like Wikimedia Commons Wikidata, has made it easy for artificial intelligence models to take advantage of the base of huge knowledge.

Wikimedia Deutschend, the German chapter of the organization, has released a new resource called Wikidata project. It takes approximately 120 million open data points stored in Wikidata and turns it into simpler format for large language models for already use.

Although the structured data of Wikidata is already readable, it was not directly compatible with obstetric artificial intelligence systems, which were designed to work with the natural language.

The new project translates Wikidata into vectors, which are numerical coordinates in the first place that show the extent of the different data related to each other.

Think about it like a map in which the terms associated closely related to the “dog” and “puppy” together, while these are not relevant such as “Dog” and “Bank Account” are very far apart. This artificial intelligence systems help understand the terms in context and address them more effectively in the natural language.

Wikimedia Deutsalland said in A. press release. She said that most of the artificial intelligence systems are currently dependent on non -dark data groups.

The secondary goal is to settle the stadium. By providing Wikidata, Wikimedia says it hopes that the smaller artificial intelligence companies can compete with technology giants that can enjoy resources to overcome the same data.

“The launch of the inclusion project is that the strong Amnesty International should not be controlled by a handful of companies – it can be developed publicly and cooperatively,” said Felipe Sadi, director of the Wikidata AI project in a statement.

Wikimedia Deutschland has been working on the project since September 2024 in cooperation with JINA AI, which has built a system that converts Wikidata entries into vectors, and Datastax in IBM, which stores these vectors in its database.

On the other hand, the release fell only one day after the transfer of Elon Musk to X to announce that it is building a Wikipedia competitor is called Grokipedia.

“We are building Grokipedia Xai”, Mouc wrote on Tuesday. “It will be a great improvement in Wikipedia. Frankly, it is a necessary step towards the Xai goal of understanding the universe.”

Musk has repeatedly mocked Wikipedia as “Wokipedia” and complained that there is no alternative in line with more right -wing opinions. He also republished Larry Sanger, the founder of Wikipedia, who resigned in 2002 and has already tried to launch several competing projects. Sanger, a long -term critic of Wikipedia from the right, was recently published on X that Wikipedia has also become World, Academic, secular and progressive.

Musk’s attempt to build an encyclopedia stored with its favorite facts only confirms why Wikimedia launched its artificial intelligence project in the first place. As artificial intelligence continues to hire, the quality of data and the bias of these systems can affect an effect on what millions of people believe is true.