For Support: AI Joins Wikipedia

Everyone needs an AI strategy in these times, including Wikipedia. According to Wikimedia, the organization behind Wikipedia, this strategy is two-pronged and should in no way lead to people being replaced. The world's largest encyclopaedia is a constant basis for many AI models. Without this data, the models would have significantly less knowledge of the world.

Among other things, AI is intended to support the moderators. They will be provided with AI-supported workflows that "automate tedious tasks to maintain knowledge integrity." Whatever that means in concrete terms. Authors are to be given time back by being able to find information on Wikipedia more quickly –, leaving more time for "human deliberation, judgment and consensus building". AI will also help with translation, which should help with the exchange of local perspectives and contexts.

Last but not least, AI will be used in future to train new helpers. The word "scaling" should not be missing from an AI strategy, which is why Wikimedia writes on the blog post about "scaling the training of new Wikipedia volunteers through guided mentors."

In addition to a human-centered approach, Wikimedia promises to use open AI models. Apparently, no selection has been made yet.

Wikipedia is the basis of AI and wants to use it vice versa

AI chatbots are a major competitor for Wikipedia. The AI models behind them have almost all been trained on the information in the encyclopaedia. This means that they can also reproduce this knowledge. People who ask an AI chatbot may receive the answers they are looking for without having to open a Wikipedia page.

Wikipedia's business model is based on donations and voluntary work. This means that it is not threatened in the same way as the business models of publishers and other website operators. They earn money by displaying advertising on their own pages. If the click, i.e. the visit to the website, is lost because an AI chatbot has already given the answer, the advertising revenue is also lost.

"Providing freely accessible knowledge to everyone in the world is Wikipedia's mission, which has only grown in importance since the advent of generative AI. Its success is why Wikipedia is at the heart of every AI training model," writes Wikimedia. Progress is recorded on the meta-wiki page. Wikimedia only recently prepared its own data from the platform as a data set explicitly for training AI models. To do this, they worked together with Kaggle, a subsidiary of Google. Wikimedia says that this is also intended to counteract the scraping of content, which puts a strain on its own servers.

(emw)