Wikipedia's Cleanup of AI-Generated Content
spectrum.ieee.org
spectrum.ieee.org
This feels like a microcosm of software development as a whole.
There's more software than ever, yet there are fewer people engaging with said software than ever, and the good software gets buried under a giant pile of crap.
I’d say there is a lot of good software buried under good software. I think AI will help these opportunists to make better software than they would otherwise.
You can put hundreds of hours into these accounts, and people will roll the dice on a vibecoded plugin they (or a stranger on the runelite plugin repo) make with very little understanding that Jagex (the company that maintains OSRS) just sees client commands. If you look like a bot and smell like a bot, you get treated like a bot.
I now know two friends that got bans (one permanent!) for rolling their own unpublished plugins, not realizing that they asked Mod Claude for a macro set for some bossing content… oof.
And so the goalposts move, oh now we've always cared about user resources, AI is using too much, but in 3 months when it's better than all human programs at resource management what will they move onto? Did we forget that unused resources are wasted resources? Or can that only justify poorly human-written programs.
I made my edits and they were immediately reverted by wikipedia's own AI bot. I gave up.
this sounds like a good safeguard
The only way to win is not to play.
> Yesterday, AI was something to clean up. Today, it’s a tool to help us with the cleanup. Tomorrow, it will be the air we breathe.
https://wikiedu.org/blog/2026/01/29/generative-ai-and-wikipe...
https://wikiedu.org/blog/2026/07/14/generative-ai-and-wikipe...
"Low-background Steel: content without AI contamination (jgc.org)" 10.jun.2025 https://news.ycombinator.com/item?id=44239481 269 comments
"Wikipedia: Database Download (wikipedia.org)" 27.apr.2025 https://news.ycombinator.com/item?id=43811732 99 comments
"Low Background Steel (lowbackgroundsteel.ai)" 27.apr.2025 https://news.ycombinator.com/item?id=43814531 2 comments
It feels like the distribution of sources/links it outputs is 33% Reddit, 33% Random blog, 33% Not Found.
Users on Reddit say…
- ADL report finds 91% of banned editor’s Israel-related edits still live on key pages. https://www.timesofisrael.com/biased-wikipedia-edits-on-isra...
- Forensic details of Oct. 7 atrocities (sexual violence, burned victims) removed and never restored on Gaza war page. https://www.adl.org/resources/press-release/adl-wikipedias-d...
- Editor JDiala admits 10-15% of active accounts since Oct. 7 were sockpuppets or Discord-coordinated for pro-Palestine bias. https://www.jpost.com/opinion/article-911291
- Hamas page restructured days after Oct. 7 to move antisemitism and abuses material to low-traffic “Criticism” article. https://www.foxnews.com/media/first-fox-wikipedias-editorial...
- New cohort of 19 editors uses subtle methods (undoing, hedging, slow rewrites) and controls latest versions on ~60% of examined pages. https://www.jpost.com/diaspora/antisemitism/article-911852
Most of its non-neutral articles are biased and don't say they are biased. The only articles that are decent are math or astrophysics, anything related to news or current events always leans one way.
I was afraid of that. I grabbed a 2024-04-20 copy as a backup, in case it turned to slop, but I suspected it might be tainted already.
Still, 2023-10 is probably the last pure https://en.wikipedia.org/wiki/Low-background_steel
A year ago somebody make a page for wikitok, my 15 minutes of fame website. About a year after that, the user was banned for mass creating articles, allegedly. The article was then nominated to be deleted, because of the nature of the creator being banned for AI generated slop. A little while later, the article was deleted just because nobody bothered to vouch for it, funny times.
When you think of something in wikipedia, you may associate it with longevity, archival... Not the case, just like the web, links rot, things change, people play games / politics. It's all the same!
Now it's pages and pages and pages of text: probably factual, but not a satisfying read. It feels like it's for a different audience, and it's not satisfying to read.
Is it AI generated? I don't know for sure, but it feels like either someone (or something)'s missing the point of who Wikipedia's audience is; or that someone is following an agenda, (or that there's an edit war going on between dueling agendas and the facts are added to drown out the "AIDS is a consequence of bad morals" agenda.)
A digital encyclopedia has different constraints, and the hard part was always trying to figure out what to leave out. Also in encyclopedias, you get what you get from the one or two people that worked on that particular article. If they had biases it would shape what everyone in the world saw. In an online article it can become a battle on what people want to see.
It is a difficult problem.
There's a pretty stark difference between the first few paragraphs of the HIV/AIDS page and the first few paragraphs of the Type 2 Diabetes page.
Both are "factual", but when I read the HIV/AIDS page, I feel like someone is trying to convince me "HIV is not a bad disease".
Obviously this is subjective. I'm not saying you're wrong, just genuinely curious at what made you think this. I'm not making assumptions, but perhaps you have (unfortunately) more personal and upsetting experiences with HIV/AIDS than with diabetes? I know that when I used to bartend at a neighborhood gay bar frequented by gen Xers it was very difficult to see and hear people lose or talk about losing their partners to something that alienated them even more than they already are/were.
It is interesting to think about how to convey information about these things though. I think it's probably a good thing to make HIV/AIDS not sound so terrifying. For pretty much every adult alive today, I'd assume, their idea of HIV/AIDS was something like being told it's a "gay disease" or that you can't touch people with it (thank you for your service, princess Diana) or some sort of anti-African racist nonsense. So conveying accurate information without bigotry, homophobia, or xenophobia is probably a good thing overall.
Also curious what sort of possible conspiracy would benefit from Wikipedia readers from not being more afraid of HIV/AIDS.
It feels like the first 2 paragraphs are doing a lot of work to downplay the seriousness of the illness:
- highlighting the "undetectable = untransmittable" PR campaign.
- reassurance that you can live a normal life with HIV, "die with virus, not of virus"
- reassurance that there's zero transmission risk with long term treatment
- reassurance that HIV is a "manageable chronic health condition"
After the first 2 paragraphs of the HIV article, I'm left thinking "okay, this isn't as bad as I thought" -- but maybe that's okay since historically there has been a lot of misinformation around HIV.
> what sort of possible conspiracy would benefit from Wikipedia readers from not being more afraid of HIV/AIDS.
De-stigmatization of HIV is the agenda. (An agenda I support)
Always fun to think about different viewpoints. The same factual information could be represented (and is, as far as I know) in another way to denounce homosexuality and premarital sex, for example. Not making a statement about any particular belief system.
Thanks for the engaging think on a difficult night for me. I appreciate it. :)
Hope your night works out well in the end.
To bring it back to the topic, this is certainly a serious concern with AI-generated content. The world changes and evolves and things that are acceptable become unacceptable and vice versa. With training data potentially being stuck in time due to the majority of new data being generated, history could essentially just stop here. Or at least that is an idea that's been proposed. I'll be honest that I have thought that to be a genuine concern. However, objectively, novel ideas have come out of these new technologies. So it's a bit of a pickle to figure out how that relates to the human experience as time marches on.
Thank you. The world will keep on spinning. :)
Also, after you had unsafe sex or accidentally touched infected injection, you can get early anti-hiv drugs - which means you will not get HIV.
I mean, what exact bias is there?
“An article should not be a complete presentation of all possible details, but a summary of accepted knowledge regarding its subject“
https://en.wikipedia.org/wiki/Wikipedia:What_Wikipedia_is_no...
https://en.wikipedia.org/wiki/History_of_HIV/AIDS
On reflex, I searched for "reagan". 0 results as of the morning of October 27, 2026.
I was there in the 1980s when the AIDS outbreak started getting mainstream news coverage. I watched the Ryan White movie on TV, when it was the biggest topic at school for months afterwards:
https://en.wikipedia.org/wiki/Ryan_White
I can tell you what really happened. HIV/AIDS scared the bejesus out of the political establishment of the time. Who then stoked that fear to push their socially conservative agenda onto everyone in an attempt to return the US to what they viewed as a morally just culture (basically the 1950s), much like MAGA today.
And just like when Donald Trump refused to wear a mask to set an example in protecting the elderly and other vulnerable groups from infection by COVID-19, Ronald Reagan refused to devote real resources to slowing the spread of of AIDS and finding a cure. Because AIDS furthered the GOP's efforts to expand the War on Drugs, push abstinence-only sex education, demonize same-sex relationships and criticize secular culture in general.
In a very real sense, Ronald Reagan and his administration let AIDS grow from being a manageable disease to becoming an epidemic that killed millions of people. His misplaced priorities set back research and treatment options by decades.
Omitting Reagan's name from the history of AIDS is like casually leaving Benedict Arnold out of the history of the American Revolutionary War.
The domino effect of his abdication of duty has touched everything from US police interventions in developing nations that it's exploited to its $40 trillion national debt. It's literally fogged the lens through which we see the world.
Its easy to forget all of that now that once-daily HIV treatment pills and bi-monthly preventative injections exist, and we're finally seeing cures on the horizon.
Revisionist history now dominates our political discourse. I write this because the "AIDS is a consequence of bad morals" agenda is not something to be lightly swept under the rug. It's literally a pillar upon which the alternate timeline we're on now stands.
I remember Reagan manufacturing AIDS into some sort of plague sent by God to punish brown skinned alongside gay people. I'm not surprised that historians have whitewashed Reagan's racism.
It was Reagan's administration that got the Fairness Doctrine repealed. That single action made the right wing media possible. Fox could never have existed prior.
The HPV vaccine prevents several cancers. Australia, which has mandatory HPV vaccination, is seeing a 95% drop in the cervical cancers that the vaccine prevents compared with pre-vaccine people. In my state, as well as some other red states, want the HPV shot banned or at the very minimum requiring positive parental consent.
And while I agree it's too long, do note that about 50% is the References section.
What I think is interesting is that in this day and age of AI with free Open Source web server software, unbelievably cheap data storage, and very inexpensive Internet hosting prices,
basically anyone can run their own version of Wikipedia
So, do specific Wikipedia articles lean too far to the left or too far to the right for your tastes?
Do specific Wikipedia articles lean too liberal or too conservative for you?
Do specific Wikipedia articles lean too pro-Democrat or pro-Republican for you?
Maybe you're just a Scientist or Physicist or Electrical Engineer and you don't like the censorship of "Free Energy" articles, er, I mean "Overunity", er, I mean "Cold Fusion", er, I mean "Low Energy Neutron Reactions" (LENR), er, I mean, "Free energy can't possibly exist, and we know this to be absolutely true!"...
Maybe you're a Programmer or Sysadmin, and you have your own opinions on the "systemd is not good" vs. "systemd is good" debate!
Maybe you're a Mathematician and you feel that some of the Wikipedia Math articles are either too complex for anyone to understand them effectively -- or conversely, too simple, and omit important facts!
Heck, maybe you think Chunky is a Soup, but then again, maybe you think it "eats like a meal!" and the associated Wikipedia page doesn't accurately reflect this belief!
Maybe you think New Shimmer is desert topping! But then again, maybe you think it's a floor wax! But you believe that the associated Wikipedia page doesn't present these facts accurately or fairly, in a non-biased way!
Well, there's a simple solution to all of these problems, all of these dilemmas, and more! Which is:
Basically run your own version of Wikipedia!
AI can be used en masse, to clean up the content of any Wikipedia web pages you don't like!
Simply fork Wikipedia, grab copies of its data (which is or should be public domain), write a prompt to look for and change (according to your own political, relgious, moral, or just-plain-common-sense beliefs!) all of the Wikipedia web pages that you feel are not "the truth, the whole truth, and nothing but the truth, so help [them/me/you/they] G-d!", and then host the thing yourself!
It's not impossible in this AI age!
Note that I am not suggesting that everyone do this; I am merely alluding to the fact that with AI, this task is now within the realm of possibility to be accomplished, whereas previously, it would have been significantly difficult...
Anyway, great article!
(And for the record -- Wikipedia is usually about 98%-99% correct on most hard STEM subjects, and an excellent starting point for research, so I'm not saying all of the above to knock Wikipedia in any way! We should all remember that Wikipedia was revolutionary for its time, and remains an excellent starting point for STEM research!)
Of course I'd love for this experiment to be done outside of the main site, and only integrated when it's mature, stable, and proven to be actually useful, but other than that implementation concern, I feel this is something that needs to be explored.
https://www.theguardian.com/uk-news/2020/aug/26/shock-an-aw-...
If it’s that easy for me to correct it, it should be automated.
As a consequence I am much happier using and adding and contributing to the english wikipedia than to the german wikipedia.
Not just random output of who knows what model, from who knows what prompt.
I wouldn't even have much of a problem with it - if these labs would be fair enough to humanity as a whole to provide the raw scans after the process is completed, and if they would limit themselves to books that have been on the market for more than 20 years.
That does not really make it a law though.
https://en.wikipedia.org/wiki/Stripped_book
> A stripped book is a mass market paperback that has been stripped of its cover in order to be recycled. The covers are returned to the publisher as evidence that the books have been destroyed and to obtain a credit on the purchase price.[1] The books are meant to be destroyed by pulping but some find their way to street merchants, sometimes by illegal means, and are resold.[2]
I grew up reading books with this ditty in them:
> If you purchased this book without a cover you should be aware that this book is stolen property. It was reported as “unsold and destroyed” to the publisher and neither the author nor the publisher has received any payment for this “stripped book.”
Those books certainly weren't scanned before being pulped. Would've been illegal. Really, the big thing that destroys books is the copyright term being as long as it is.
What was going on recently was AI companies were specifically seeking out rare — or sometimes irreplaceable — books to supplement their data sets (and hopefully getting a leg up on their competitors). The destruction of the books wasn't the point, it was just a consequence of the business-oriented realization that they could scan the contents of more books faster if they just chopped off the bindings rather than investing the time to care for the books and keep them intact (plus supposedly playing along with copyright law by "only having one copy of the copyrighted material").
A lot of worthless books are rare. That was what people in the trade call "Clickbait"; in that it isn't technically a lie but it sure ain't the truth.
> The destruction of the books wasn't the point
It was to the people who shared the stories for clicks and outrage farming.
Unfortunately, this would get them an even bigger copyright smackdown than they've already received.
That is AI is a tool. I've driven screws with a hammer - it worked in a pinch, but it was not good. People have also used hammers to kill others.
That takes a lot of commitment, you really have to want that screw in pretty badly! And when you're done it ends up just being a worse nail.
I've seen these used to assemble wooden palettes (they are VERY hard to remove). A search on Home Depot reveals they may be used for certain types of siding as well.
I built a house with them. You really want to check twice if you're putting them in where you want them because getting them out is - as you noted - annoyingly difficult. But they're very nice for large surfaces that are going be battered by the wind, they're much longer than your typical screw and working with them is very fast.
I absolutely understand where you are coming from, and I agree to a point. If you're of an age where you actually read a printed encyclopedia, the tone and voice of the entire set was sameish as well. The difference is that it didn't sound LLMish like LLM text does. I know that printed encyclopedias would have been vetted by editors, and I feel this is the key difference. Editors not only knew grammar rules, but had that sense of good writing practices that's hard to write down as rules. A piece of text passed through a good editorial review will read human while LLM text reads as LLM text.
When we say AI sounds same-ish we mean all sources of content can sound the same - fiction and non-fiction, news articles and blog posts, marketing pitches and ai drafted personal messages, all sound kinda the same. That is a different situation, and that's before you get to "and that one sound is LLMish".
Having a program constantly suggest "better" ways of phrasing things is stripping your personality from your language. Maybe this is needed if you are writing corporate PR pieces, but you should avoid it for anything else.
Of course AI can be used to check sources, get critical feedback and translate to languages. But content should be 100% human driven.
So the answer to "Why do we assume AI-generated means slop?" is probably "experience", and any proposition starting with "If the content is good enough" is trivially true as a conditional statement with a false antecedent (i.e., ⊨(⊥⇒Q) for all Q).
It's a pretty short article compared to what usually gets posted here.
What then, is the problem of AI-Generated Content in Wikipedia?
> Fake or misleading citations have become a major worry.
Why is this not a standalone problem? Humans can fake and mislead in citations too, in fact, this is a major problem and has been so, for literally centuries in science.
> Being a place that was, in theory, free from AI slop has become part of Wikipedia’s identity.
Isn't this a separate issue in itself too? One of "quality". If AI generates poor quality content, than I would think the poor quality is the problem, not the producer of it.
The article mentions several times how hard it is to detect AI content. If it is that hard, then obviously AI is doing a good job and producing content that matches that of humans. Why then insist on one type of producer?
Yes, but who will review the mountains of slop to determine whether it is of acceptable quality? The main problem AI introduces is high volume at low quality, often overwhelming the human resource bottlenecks these systems have in place.
Apparently Wikipedia has good reasons to not change this process fundamentally, other than add extra filters at the start to try and detect low quality posts, based on attributes that show a correlation with low quality.
Sounds like a pragmatic, but far from ideal solution to me. Correlation is not causation.
Today is not the day. Especially not regarding the verification.
I would rather be worried about the quality of all the primary and secondary sources going forward.
What do you mean by this? I have passed very large documents into LLMs and they are able to accurately extract quotes and identify their location in the source documents. I don’t see any reason to think that they couldn’t accurately quote from and cite existing literature.
There are tools where they try to retroactively add references, the LLM interfaces try to do this, but it's usually inaccurate.
I think of the way a scientific article works. "The sky is blue (Pearson, 1992)." The writer is fully aware what the factual statement is and which publication supports it (not just that, but which kind of publication - primary, secondary source, peer reviewed?). An LLM can try to imitate this, but the actual information to support this does not exist in its training.
'Knowledge' formed this way is a living, evolving graph, and sometimes links are broken (findings are reevaluated or disproven); having the content declaration so clearly in the open makes it easier to deal with this evolving landscape.
The hallucination era is rapidly ending.
Is 80/20 ever generally correct? Why is it favored over the 90/10 meme? Why would it apply to this? Why wouldn’t the numbers be vastly different for different domains of content generation? Etc.
If you pay attention, so much user content is just repetition of something they read on the same fucking website and uncritically checked into their brain.
Current HN top post is Roblox downloaded 145GB models from us (including adult content) without asking. I fully expect to start seeing comments to the tune of “Roblox is stealing porno to train AI for children” within the day in any discussion Roblox comes up in.
Why would these folk’s contribution to any other user content site be any more thoughtful?
It’s just a rough way to say “mostly this with a decent bit of that” because 20% seems like a medium-small percent that could or could not matter depending on the context
That’s all anyone means when they say 80/20
To your second point, the 80% could be lots of basic things and the 20% could be the important facts, so it’s not always meant as a dismissal that the 20 doesn’t matter
Sometimes. But it, or 90/10, or sometimes 75/25, is understood to main “a large part of the time”/“a significantly smaller but still crucial amount off the time”.
> If you pay attention, so much user content is just repetition of something they read on the same fucking website and uncritically checked into their brain.
Which is exactly the problem. We now have LLMs doing the same thing, flooding everywhere with more of that noise, further hiding what good signal remains. Even if some of that good signal is by chance almost entirely LLM generated too, it appears to be at best a vanishingly small amount.
Funny though, as that admission of error makes it more reliable than if it tried to hide this fact.
I have access to a number of LLMs and can order it to build me a wiki site with all knowledge of the world, just like everyone else with the same access. What would make mine any more valuable or better than another?
Latentpedia starts from wikidata, not Wikipedia.
Even subtle bias can manifest in significant ways. Given that the predominant market leaders in producing such models are currently : A country that's attempting to squeeze the word "america" in any geographic formation possible + rewriting history to put itself first, and a country that has perfected censorship en-mass and already actively doctors its own history.
I'm not convinced giving LLMs carte blanche to write authoritative sources on topics that - often are ideologically sensitive/controversial isn't just needlessly reckless and short-sighted. It's not a far reach to imagine such models reflecting the current views of administrations in these countries and becoming reluctant to mention certain colonial ambitions, environmental damage, political corruption, etc. where it might be relevant to authoritative content.
Wikipedia already struggles with issues in idealogical bias from authors - human validation (somewhat)helps mitigate that. You could argue that human validation is still prone to same bias - but you can always draw on a more diverse pool of editors with little ideological overlap. With LLM validation you're pretty much at the mercy of a pretty incestuous pool of data & a handful of self-serving corporates & governments.
As a result, I don't think LLMs will reach this state easily.