I think readers take for granted how concise Wikipedia's prose tends to be. AI, in comparison, seems built to ramble, being overly specific where it doesn't need to be and lacking specificity where it ought to have it.
When you think about it, "what should go on a thing's Wikipedia page?" is an interesting question; the answer certainly isn't "anything and everything." AI just doesn't have a good sense for what belongs, I feel.
I don’t do this systematically, just sometimes out of curiosity.
But it is always the same pattern: bloat, bloat, bloat.
What I very critically witness is the so called gender neutrality movement where large bodies of text get rewritten to fulfill a political agenda.
This is a major loss of quality. Hundreds of years of using language and getting results by using it as a means and if you compare recent downfalls in connection with gender politics you should be very worried of not already.
Even if some admins drive such agendas, why not use a new mode like a new language for those who want it? This would have been the old skill Wikipedia way and the actual edit wars that aren’t sadly made Wikipedia lose massive credibility for me.
Can't wait for the 80 page Talk threads.
It seems a smaller "win" than most think. Just discourages wholesale rewriting and creation of new articles using AI. Assistance with editing is explicitly allowed.
> Text generated by large language models (LLMs) such as ChatGPT, Gemini, Claude, DeepSeek etc. often violates several of Wikipedia's core content policies. For this reason, the use of LLMs to generate or rewrite article content is prohibited, save for these two exceptions
(The two exceptions are basic copyediting and translation).
I don't see how this is unintelligent and impractical. Wikipedia are trying to protect their core content policies ie. the very things that separate Wikipedia from Conservapedia, Grokipedia, RationalWiki or any other wiki. They are willing to grant exceptions in cases where LLMs are valuable.
And they even acknowledge that:
> Some editors may have similar writing styles to LLMs. More evidence than just stylistic or linguistic signs is needed to justify sanctions
So it seems like the ban is only intended to be used in extremely egregious cases.
AI generated articles are, on the balance, inferior, except for people that want simple, low quality content.
But LLMs are moving up the value chain with Deep Research. They can give explanations tuned to a reader's knowledge/viewpoints and provide interactive content Wikipedia doesn't support. That is a killer app for math/science topics.
Wikipedia will win against a generic corporate encyclopedia on neutrality/oversight, but it'll lose badly on UX, which is what matters.
I think the tipping point will be direct integration of academic sources into ChatGPT/Claude/Gemini and a "WikiLink" type way to discover interesting follow-up topics.
I can't trust AI answers for serious historical or social science topics because of the first. And generally my chat with AI ends once I get the answer I need because I can't get rabbitholed into other topics.
I might be slightly wrong, but probably not by a lot, yet. Sure there's an element of "holding-it-wrong-ism" in my position. But ... it does actually take practice to get it right, and best practices are badly documented!
That said the situation is changing rapidly: https://news.ycombinator.com/item?id=47547849 "AI bug reports went from junk to legit overnight, says Linux kernel czar"
--
of course they banned ai they could barely allow css
Like, this attempt† where the bot then attempted to lecture users who were hostile towards it before it was eventually banned.
† https://en.wikipedia.org/wiki/User_talk:TomWikiAssist