Jimmy Wales Says Wikipedia Could Use AI. Editors Call It the 'Antithesis of Wikipedia'
Jimmy Wales Says Wikipedia Could Use AI. Editors Call It the 'Antithesis of Wikipedia'
Wales’s quote isn’t nearly as bad as the byline makes it out to be:
Wales explains that the article was originally rejected several years ago, then someone tried to improve it, resubmitted it, and got the same exact template rejection again.
“It's a form letter response that might as well be ‘Computer says no’ (that article's worth a read if you don't know the expression),” Wales said. “It wasn't a computer who says no, but a human using AFCH, a helper script [...] In order to try to help, I personally felt at a loss. I am not sure what the rejection referred to specifically. So I fed the page to ChatGPT to ask for advice. And I got what seems to me to be pretty good. And so I'm wondering if we might start to think about how a tool like AFCH might be improved so that instead of a generic template, a new editor gets actual advice. It would be better, obviously, if we had lovingly crafted human responses to every situation like this, but we all know that the volunteers who are dealing with a high volume of various situations can't reasonably have time to do it. The templates are helpful - an AI-written note could be even more helpful.”
That being said, it still reeks of “CEO Speak.” And trying to find a place to shove AI in.
More NLP could absolutely be useful to Wikipedia, especially for flagging spam and malicious edits for human editors to review. This is an excellent task for dirt cheap, small and open models, where an error rate isn’t super important. Cost, volume, and reducing stress on precious human editors is. It's a existential issue that needs work.
…Using an expensive, proprietary API to give error prone yet “pretty good” sounding suggestions to new editors is not.
Wasting dev time trying to make it work is not.
This is the problem. Not natural language processing itself, but the seemingly contagious compulsion among executives to find some place to shove it when the technical extent of their knowledge is occasionally typing something into ChatGPT.
It’s okay for them to not really understand it.
It’s not okay to push it differently than other technology because “AI” is somehow super special and trendy.
This is another reason why I hate bubbles. There is something potentially useful in here. It needs to be considered very carefully. However, it gets to a point where everyone's kneejerk reaction is that it's bad.
I can't even say that people are wrong for feeling that way. The AI bubble has affected our economy and lives in a multitude of ways that go far beyond any reasonable use. I don't blame anyone for saying "everything under this is bad, period". The reasonable uses of it are so buried in shit that I don't expect people to even bother trying to reach into that muck to clean it off.
So... I actually proposed a use case for NLP and LLMs in 2017. I don't actually know if it was used.
But the usecase was generating large sets of fake data that looked real enough for performance testing enterprise sized data transformations. That way we could skip a large portion of the risk associated with using actual customer data. We wouldn't have to generate the data beforehand, we could validate logic with it, and we could just plop it in the replica non-prodiction environment.
At the time we didn't have any LLMs. So it didn't go anywhere. But it's always funny when I see all this "LLMs can do x" because I always think about how my proposal was to use it... For fake data.
This bubble's hate is pretty front-loaded though.
Dotcom was, well, a useful thing. I guess valuations were nuts, but it looks like the hate was mostly in the enshittified aftermath that would come.
Crypto is a series of bubbles trying to prop up flavored pyramid schemes for a neat niche concept, but people largely figured that out after they popped. And it's not as attention grabbing as AI.
Machine Learning is a long running, useful field, but ever since ChatGPT caught investors eyes, the cart has felt so far ahead of the horse. The hate started, and got polarized, waaay before the bubble popping.
...In other words, AI hate almost feels more political than bubble fueled. If that makes any sense. It is a bubble, but the extreme hate would still be there even if it wasn't.
That being said, it still wreaks of “CEO Speak.”
I think you mean reeks, which means to stink, having a foul odor.
Those homophones have reeked havoc for too long!
Waves hands "You didn't see anything."
Thank you. Glad to know I am not the only one that got triggered, lol.
That being said, it still wreaks of “CEO Speak.” And trying to find a place to shove AI in.
I don't see how this is "shoved in." Wales identified a situation where Wikipedia's existing non-AI process doesn't work well and then realized that adding AI assistance could improve it.
Neither did Wales. Hence, the next part of the article:
For example, the response suggested the article cite a source that isn’t included in the draft article, and rely on Harvard Business School press releases for other citations, despite Wikipedia policies explicitly defining press releases as non-independent sources that cannot help prove notability, a basic requirement for Wikipedia articles.
Editors also found that the ChatGPT-generated response Wales shared “has no idea what the difference between” some of these basic Wikipedia policies, like notability (WP:N), verifiability (WP:V), and properly representing minority and more widely held views on subjects in an article (WP:WEIGHT).
“Something to take into consideration is how newcomers will interpret those answers. If they believe the LLM advice accurately reflects our policies, and it is wrong/inaccurate even 5% of the time, they will learn a skewed version of our policies and might reproduce the unhelpful advice on other pages,” one editor said.
It doesn't mean the original process isn't problematic, or can't be helpfully augmented with some kind of LLM-generated supplement. But this is like a poster child of a troublesome AI implementation: where a general purpose LLM needs understanding of context it isn't presented (but the reader assumes it has), where hallucinations have knock-on effects, and where even the founder/CEO of Wikipedia seemingly missed such errors.
Don't mistake me for being blanket anti-AI, clearly it's a tool Wikipedia can use. But the scope has to be narrow, and the problem specific.
Why is leadership always so vapid and disconnected from reality?
Because that's what being in a position of power does to a mf
Because this is one of the rare times he sat down at the keyboard to do the real work being done by people in this organization and he realized that it’s hard and he wants a shortcut. He sees his time as more valuable and sees this task as wasting his time, but it is their primary task and one they do as volunteers because they are passionate about it. He’s not going to get a lot of traction with them telling them the thing they do for free because they love it isn’t worth anyone’s time.
I think commenters here don't actually do Wikipedia. Wales was instrumental in Wikipedia's principles and organization besides the first year of Sanger. He handpicked the first administrators to make sure the project would continue its anarchistic roganization and prevent a hierarchy from having a bigger say in content matters.
I would characterize Wales as a long-retired leader rather than leadership.
Remember you can download all of Wikipedia in your language and safely store it on a drive buried in your backyard, for after they rewrite history and eliminate freedom of speech.
Already got it downloaded. It's only like 100 - 150 gigabytes or something like that. Got it on my PC, my laptop, and my external hard drive. I don't trust the powers that be to keep it intact anymore so I'd rather have my own copy, even if outdated.
What about any of this remotely connects to "rewriting history and eliminating freedom of speech?"
Proprietary AI means corpo involvement, and usually it's the really actively awful sort of techbros, this involvement gives them some power, and this power is a threat. Whether it materializes or not, living in the world we do now, it's only right to be wary. I already figured Wikipedia was on its way out a few months ago and downloaded both the kiwi program reader version and the raw xml dump + file for truly apocalyptic situations.
Some people can't really stop seeing conspiracies everywhere.
They're trying to get rid of Wikipedia by saying they're shit and doing things you'll hate. Fight for no AI if that's your thing, but read very carefully what's happening. Wikipedia can NOT go away.
The problem with LLMs and other generative AI is that they're not completely useless. People's jobs are on the line much of the time, so it would really help if they were completely useless, but they're not. Generative AI is certainly not as good as its proponents claim, and critically, when it fucks up, it can be extremely hard for a human to tell, which eats away a lot of their benefits, but they're not completely useless. For the most basic example, give an LLM a block of text and ask it how to improve grammar or to make a point clearer, and then compare the AI generated result with the original, and take whatever parts you think the AI improved.
Everybody knows this, but we're all pretending it's not the case because we're caring people who don't want the world to be drowned in AI hallucinations, we don't want to have the world taken over by confidence tricksters who just fake everything with AI, and we don't want people to lose their jobs. But sometimes, we are so busy pretending that AI is completely useless that we forget that it actually isn't completely useless. The reason they're so dangerous is that they're not completely useless.
It’s almost as if nuance and context matters.
How much energy does a human use to write a Wikipedia article? Now also measure the accuracy and completeness of the article.
Now do the same for AI.
Objective metrics are what is missing, because much of what we hear is “phd-level inference” and it’s still just a statistical, probabilistic generator.
https://www.pcmag.com/news/with-gpt-5-openai-promises-access-to-phd-level-ai-expertise
It is completely useless as presented by the major players who atrocities trying to jam models that are trying to everything at the same time and that is what we always talk about when discussing AI.
We aren't talking about focused implementations that are Wikipedia to a certain set of data or designed for specific purposes. That is why we don't need nuance, although the reminder that we aren't talking about smaller scale AI used by humans as tools is nice once in a while.
ok jimmy boy, will that ai also beg for donations? /s
"Editors" are not a unified block. I would be fine with it, depending on how it's used.
if jimmy wales puts ai in wikipedia i stg imma scream