Hey everyone, this is Olga, the product manager for the summary feature again. Thank you all for engaging so deeply with this discussion and sharing your thoughts so far.
Reading through the comments, it’s clear we could have done a better job introducing this idea and opening up the conversation here on VPT back in March. As internet usage changes over time, we are trying to discover new ways to help new generations learn from Wikipedia to sustain our movement into the future. In consequence, we need to figure out how we can experiment in safe ways that are appropriate for readers and the Wikimedia community. Looking back, we realize the next step with this message should have been to provide more of that context for you all and to make the space for folks to engage further. With that in mind, we’d like to take a step back so we have more time to talk through things properly. We’re still in the very early stages of thinking about a feature like this, so this is actually a really good time for us to discuss here.
A few important things to start with:
- Bringing generative AI into the Wikipedia reading experience is a serious set of decisions, with important implications, and we intend to treat it as such.
- We do not have any plans for bringing a summary feature to the wikis without editor involvement. An editor moderation workflow is required under any circumstances, both for this idea, as well as any future idea around AI summarized or adapted content.
- With all this in mind, we’ll pause the launch of the experiment so that we can focus on this discussion first and determine next steps together.
We’ve also started putting together some context around the main points brought up through the conversation so far, and will follow-up with that in separate messages so we can discuss further.
How about not putting AI into something that should be entirely human controlled?
Yeah as more organizations implement LLMs Wikipedia has the opportunity to become more reliable and authoritative. Don’t mess that opportunity up with “AI.”
These days, most companies that work with web based products are under pressure from upper management to “use AI”, as there’s a fear of missing out if they don’t. Now, management doesn’t necessarily have any idea what they should use it for, so they leave that to product managers and such. They don’t have any idea, either, and so they look at what features others have built and find a way to adapt one or more of those to fit their own products.
Slap on back, job well done, clueless upper management happy, even though money and time have been spent and the revenue remains the same.
deleted by creator
I’ve already posted this a few times, but Ed Zitron wrote a long article about what he calls “Business Idiots”. Basically, people in decision making positions who are out of touch with their users and their products. They make bad decisions, and that’s a big factor in why everything kind of sucks now.
https://www.wheresyoured.at/the-era-of-the-business-idiot/ (it’s long)
I think a lot of us have this illusion that higher ranking people are smarter, more visionary, or whatever. But I think no. I think a lot of people are just kind of stupid, surrounded by other stupid people, cushioned from real, personal, consequences. On top of that, for many enterprises, the incentives don’t line up with the users. At least wikipedia isn’t profit driven, but you can probably think of some things you’ve used that got more annoying with updates. Like google putting more ads up top, or any website that does a redesign that yields more ad space, worse navigation.
The sad truth is that AI empowers the malicious to create a bigger impact on workload and standards than is scalable with humans alone. An AI running triage on article changes that flags or reports changes which need more input would be ideal. But threat mitigation and integrity preservation don’t really seem to be high on their priorities.
deleted by creator
Articles already have a summary at the top due to the page format, why was AI shoved into the process?
Because AI
I can’t wait until this “put LLMs in everything” phase is over.
So they:
- Didn’t ask editors/users
- noticed loud and overwhelmingly negative feedback
- “paused” the program
They still don’t get it. There’s very little practical use for LLMs in general, and certainly not in scholastic spaces. The content is all user-generated anyway, so what’s even the point? It’s not saving them any money.
Also it seems like a giant waste of resources for a company that constantly runs giant banners asking for money and claiming to basically be on there verge of closing up every time you visit their site.
I also think that generating blob summaries just goes towards brain rot things we see everywhere on the web that’s just destroying people’s attention spam. Wikipedia is kind of good to read something that is long enough and not just some quick, simplistic and brain rotting inducing blob
If her list were straight talk:
- Were gonna make up shit
- But don’t worry we’ll manually label it what could go wrong
- Dang no one was fooled let’s figure out a different way to pollute everything with alternative facts
deleted by creator
Since much (so-called) “AI” basic training data depends on Wikipedia, wouldn’t this create a feedback loop that could quickly degenerate ?
Yes
Only if the summary is included in the training data.
Lol, the source data for all AI is starting to use AI to summarize.
Have you ever tried to zip a zipfile?
But then on the other hand, as compilers become better, they become more efficient at compiling their own source code…
Yeah but the compilers compile improved versions. Like, if you manually curated the summaries to be even better, then fed it to AI to produce a new summary you also curate… you’ll end up with a carefully hand-trained LLM.
So if the AI generated summaries are better than man made summaries, this would not be an issue would it?
If AI constantly refined its own output, sure, unless it hits a wall eventually or starts spewing bullshit because of some quirk of training. But I doubt it could learn to summarise better without external input, just like a compiler won’t produce a more optimised version of itself without human development work.
I don’t see how AI could benefit wikipedia. Just the power consumption alone isn’t worth it. Wiki is one of the rare AI free zones, which is a reason why it is good
If they thought this would be well-received they wouldn’t have sprung it on people. The fact that they’re only “pausing the launch of the experiment” means they’re going to do it again once the backlash has subsided.
RIP Wikipedia, it was a fun 24 years.
Not everything is black and white, you know. Just because they have this blunder, doesn’t mean they’re down for good. The fact they’re willing to listen to feedback, whatever their reason was, still shows some good sign.
Also keep in mind the organization than runs it has a lot of people, each with their own agenda, some with bad ones but extremely useful.
I mean yeah, sure, do ‘leave’ Wikipedia if you want. I’m curious to where you’d go.
Me saying “RIP” was an attempt at hyperbole. That being said, shoehorning AI into something for which a big selling point is that it’s user-made is a gigantic misstep - Maybe they’ll listen to everybody, but given that they tried it at all, I can’t see them properly backing down. Especially when it was worded as “pausing” the experiment.
the fact they’re willing to listen to feedback, whatever their reason was, is a good sign
Oh you have so much to learn about companies fucking their users over if you think this is the end of them trying to shove AI into Wikipedia
Then teach me daddy~
I canceled my recurring over this about a week ago, explaining that this was the reason. One of their people sent me a lengthy response that I appreciated. Still going to wait a year before I reinstate it, hopefully they fully move on from this idea by then. It sounded a lot like this though, kinda wishy washy.
Summarization is one of the things LLMs are pretty good at. Same for the other thing where Wikipedia talked about auto-generating the “simple article” variants that are normally managed by hand to dumb down content.
But if they’re pushing these tools, they need to be pushed as handy tools for editors to consider leveraging, not forced behavior for end users.
Summaries that look good are something LLMs can do, but not summaries that actually have a higher ratio of important/unimportant than the source, nor ones that keep things accurate. That last one is super mandatory on something like an encyclopedia.
The only application I’ve kind of liked so far has been the one on Amazon that summarizes the content of the reviews. Seems relatively accurate in general.
If we need summaries, let’s let a human being write the summaries. We are already experts at writing. We love doing it.
not forced behavior for end users.
This is what I’m constantly criticizing. It’s fine to have more options, but they should be options and not mandatory.
No, having to scroll past an AI summary for every fucking article is not an ‘option.’ Having the option to hide it forever (or even better, opt-in), now that’s a real option.
I’d really love to see the opt-in/opt-out data for AI. I guarantee businesses aren’t including the option or recording data because they know it will show people don’t want it, and they have to follow the data!
Is there a way for us to complain to wikipedia about this? I contribute money every year, and I will 100% stop if they’re stomping more LLM-slop down my throat.
Edit: You can contribute to the discussion in the link, and you can email them at addresses found here: https://wikimediafoundation.org/about/contact/
It does sound like it could be handy
I don’t think Wikipedia is for the benefit of users anymore, what even are the alternatives? Leftypedia? Definitely not Britannica
Noo Wikipedia why would you do this
