Wikipedia is filled with AI-generated beliefs and fakes

Wikipedia is experiencing a crisis due to the fact that users have started publishing meaningless or unverified information generated by artificial intelligence using chatbots such as ChatGPT and the like. However, given the growing popularity of AI technologies, this was to be expected.

Image source: BoliviaInteligente/Unsplash

According to TechSpot, to solve the problem, a project called WikiProject AI Cleanup was created, which is a group of volunteers that searches, edits and removes false information supposedly added using generative AI.

Ilyas Lebleu, one of the founders of the cleanup team, said the problem came to light when Wikipedia editors and users noticed sections of articles that were clearly written by a chatbot. Suspicions were confirmed when some of these texts were recreated using ChatGPT.

«We noticed an unusual writing style that was clearly not written by a human, and we were able to reproduce these phrases using ChatGPT,” LeBleu said. — Having discovered characteristic phrases and phrases, we identified the most egregious examples of generated articles. After this, it was decided to organize a project to search for AI text.”

One example is an article about an alleged Ottoman fortress called Amberlisihar, built in the 1400s. The text, about 2,000 words long, described in detail the location and construction of this facility. However, the fortress does not actually exist, and all the information about it was completely fictitious, but looked convincing thanks to the inclusion of real facts. However, the problem concerns not only new articles on Wikipedia. Unscrupulous users insert false information into existing articles. In one case, a section dedicated to a species of crab was added to an article about a beetle, with correct links to sources.

LeBleu and his colleagues admit they don’t fully understand why people do this. However, the reasons are obvious. Firstly, this is a problem with the Wikipedia system itself, which allows anyone to become an editor. By the way, this is the reason why many universities prohibit students from using Wikipedia as their main source of information.

Secondly, it is no secret that the Internet often becomes the object of abuse, especially now that artificial intelligence has appeared. An example is the infamous Microsoft bot named Tay, which was disabled less than 24 hours after launch for posting offensive and racist tweets on X. AI is also being used to create deepfakes and books on Amazon Kindle.