Wikipedia to use AI — to augment editors, not content

The recent strategy on AI use at Wikipedia primarily focuses on making the editing and writing process easier by automating «tedious tasks.» And they won’t use AI to produce content, with one major caveat.

— The community of volunteers behind Wikipedia is the most important and unique element of Wikipedia’s success, the document begins.

It then goes on to say that AI will be used to remove technical barriers so users and editors can spend more time on the important stuff, and instead focus on AIs true strengths in automating tasks to let volunteers focus on «knowledge integrity.»

Might be used to translate
There’s one bullet point in their summary that stands out, though: Helping editors share local perspectives or context says they will use AI for automating the translation and adaptation of common topics.

So there might well be some AI slop as a result of this, with important articles showing up in hundreds of languages through machine translation, if these arent vigorously vetted. That goes doubly so for smaller language areas where articles might not get a lot of eyeballs.

Ai led to 50% traffic surge
Wikipedia goes on to say they will prioritize human agency and lean into using open source models as a matter of principle.

They recently announced that AI scrapers were swarming their site with requests, leading to a 50% surge in traffic, and even started distributing an AI friendly dataset for training on Wikipedia content.

Read more: Wikimedia’s AI strategy, story on TechCrunch, Bandwidth surge on Ars Technica and the AI dataset from Mashable.