Wikipedia declares war on AI: bans content written by artificial intelligence

The fight for the reliability of knowledge: the English Wikipedia has announced a significant policy change, according to which the use of creative artificial intelligence is prohibited for the purpose of writing or rewriting entries. The decision comes amid growing concerns about the impact of systems based on language models on the quality of information, after it became clear that they may produce content that does not meet the site’s strict editing rules.

According to the new guidelines, artificial intelligence systems sometimes tend to exceed what is required, change the meaning of texts and present information that is not backed up by sources. Because of this, Wikipedia states that they should be avoided for the purpose of creating new content or substantially editing existing entries.

However, the policy does not completely rule out the use of technology. Editors may use artificial intelligence-based tools to polish existing wording or improve style, but only on the condition that the text undergoes a strict human check to verify the facts. Relief was also given in the area of ​​translation, when automatic translation systems can be used, provided that the editor is fluent in both languages ​​and is able to identify mistakes or inaccuracies.

One of the Wikipedia community managers noted that the move was intended to encourage a broader discussion about the place of artificial intelligence in knowledge platforms. According to him, it is possible that the decision will lead to similar initiatives in other communities, which will reexamine the limits of the use of technology and its effect on the quality of information.

Along with this, it is important to emphasize that Wikipedia is not a uniform body, and each language version is run independently. This means that AI usage policies may vary between countries and languages. For example, a stricter decision has already been made on the Spanish Wikipedia, which completely prohibits any use of language models, even for drafting or translation purposes.

The main challenge that remains is enforcement. Identifying content generated by artificial intelligence is still not completely accurate, so there is a chance that such texts will continue to appear in entries, especially in those that are not under the constant supervision of active editors.