Wikipedia Co-Founder on Polarization, AI, and Elon Musk’s Grokipedia

WALTER ISAACSON: Thank you Bianna. And Jimmy Wales, thank you for joining the show.

 

JIMMY WALES: Good to be here. Thank you.

 

ISAACSON: In your book, the “Seven Rules of Trust,” you talk about creating Wikipedia way back more than two decades ago on a strange whim that everybody could group edit an encyclopedia. And one of the things that struck me is the theory of trust in the book and how that began in some ways with your daughter and her birth.

 

WALES: Yeah. So right at the beginning of Wikipedia, literally just before, my daughter Kira was born and she was very sick. She had meconium aspiration syndrome. And I, I didn’t know what that was. So I, I, you know, from the hospital, I sort of tried to, I rushed home and checked the internet, and I found some technical articles and I found some nonsense, and I just thought it was like I had been trying to create an encyclopedia project and I was nearly ready to give up, but that was when I was like, okay, I, I have to, like, I can’t give up. Like this feels like it’s really important. And that was when we decided to actually change everything and become a Wiki, meaning a website anyone can edit and abandon the old sort of top down model to really just say, okay, I’ve got this group of people and they wanna write an encyclopedia, let’s just open it up. Like let, let’s get started, let’s start writing and see what happens.

 

ISAACSON: You say abandon the old top down model you had kind of started where experts were gonna write each piece. Yeah. It seems like a crazy leap to say, wait a minute, we’ll let everybody write.

 

WALES: Well, for a couple of years we had tried a very top down system. There was a seven stage review process to get anything published. I tried to do it to write an article, and it was very intimidating and not very fun. They were gonna send my draft out to the most prestigious professors they could find, and I had realized, like, this isn’t fun. Like nobody’s really gonna do this. And so we had been talking about you know, what are some ways, and one of my employees brought me and showed me like Jeremy Rosenfeld showed me the Wiki. I was like, oh, that’s interesting. Maybe we should and then –

 

ISAACSON: The Wiki just meaning a way people can group edit some document.

 

WALES: Yeah, exactly. Exactly. The Wiki had been around for five years as a sort of underground software concept, and I was like, oh, okay, let’s try that. Let’s try to use a Wiki. And yeah, it worked. I mean, we had more work done in two weeks than we had in almost two years because we finally just let our community like get started writing and like editing each other’s work and improving it and so forth. So it was really a breakthrough for the project.

 

ISAACSON: And you say it was based on trust, but in these polarized times, where everything, whether it be wearing a mask, you know, during COVID or whatever, becomes politicized, is it harder to do a Wikified encyclopedia?

 

WALES: I mean, not really. I mean, I do think that obviously the polarization in the world makes a lot of things harder, and that’s unfortunate, the decline in trust amongst different groups of people. You know, the, the idea that, hey, you know, you and I might disagree about this, but I trust that you’re like a sensible person and we’ve got different background and therefore we’ve come to different conclusions, but we could probably find a compromise that sort of, that’s the spirit of Wikipedia and that’s the spirit that we’ve kind of lost in the world today. So it does make it harder to some extent, but, you know, there’s a lot of great people out there and there’s a lot of still – you know, I think what goes on on social media and really polarized media of all kinds doesn’t really reflect how normal people think and live. And I think most people are still, you know, perfectly happy to say, you know, oh yeah, I, I, you know, I disagree with this person politically, but you know, they’re still my friend. That’s okay.

 

ISAACSON: Yeah. But the growing number, those have pushed back on Wiki in the past year or so, ranging from people in the Republican party here, or Elon Musk, or even the person who was with you at the beginning, Larry Sanger saying, Hey, you are woke, you’re left. You skewed towards the academic. Is that true?

 

WALES: Well, skewed towards the academic might be true. I mean, certainly if you think about something, you know, medical related, we definitely prefer the New England Journal of Medicine to a tabloid newspaper.

 

ISAACSON: Yeah. But I’m talking about politics. When you have the Donald Trump entry or the Gaza entry.

 

WALES: We have a really strong commitment to neutrality, and that’s like super important. Do we always hit neutrality in the way that I would like? No, not always, but it’s always something that we strive for. And when there are criticisms about neutrality, I think the right answer for us is to say, alright, well let’s, let’s dig into the details like, what’s wrong? Like, how do we fix it? What are the sources that we’ve overlooked? Because if we don’t take that spirit of saying, actually neutrality is the most important core principle of Wikipedia, always has been. We have to double down on that in these times and say, look, it would be a huge mistake to say, well, the world’s gotten polarized, so we have to pick a side. No, the world’s gotten polarized. We need to describe that fairly. We need to sort of be thoughtful, kind, understanding, really try hard to present all the sides fairly when there’s a legitimate debate.

 

ISAACSON: Elon Musk has been one of the severest critics in some ways, what he calls “Woke-pedia,” just launched as you know Grok-opedia and it has close to a million entries now, a totally different concept. It’s not crowdsource, it’s not people writing. It uses AI to gather information from all over the internet. Tell me what you think when you looked at that.

 

WALES: Well, I haven’t had time to see it yet. It’s sort of just, it launched briefly and then it was taken down and I’m doing a book tour, so I’m doing everything. I’d say by tomorrow I’ll know more. But I haven’t had a chance really to look at it. I – people have sent me a few things. It clearly is gonna have a lot of mistakes and errors. I mean, I think one of the things that Elon may be underestimating is how hard it is to do good quality reference material and to really gather all the sources. And also in our community, we have these really long, intense debates about editorial matters. Like the exact wording of sentences is, is often like, they’re very carefully crafted by people who disagree to say like, how can we find a compromise that we can all accept as being factual? And you can’t just sort of blindly – and the thing we do know about you know, about large language models is that they can’t really be trusted. They make mistakes, they hallucinate they make stuff up out of thin air. They, they like to please you, and they are, can definitely be highly ideologically biased depending on how you’ve trained them. And so, you know, I’m a little skeptical that this could possibly work, but, you know, we’ll see.

 

ISAACSON: Do you think it’s a good idea though, to try some large language model that actually takes articles and printed things from the internet and tries to synthesize it? And might that be incorporated at some point into Wikipedia?

 

WALES: Yeah, I, I mean, I definitely think, you know, this technology is amazing and we are looking, we have a machine learning team looking into this. I’ve personally, I wrote a script that can take a short Wikipedia entry and look up all the, the, the sources at the bottom. And then I ask, is there anything in the sources that should be in Wikipedia but isn’t? Or is there anything in Wikipedia that’s not supported by the sources? Give me some suggestions. You know what, it’s pretty good. It isn’t perfect and it needs refinement, but I feel like that’s probably a path of, like, supporting the community by scanning over things and suggesting things that you can imagine, maybe two Wikipedia entries have contradictory information. Could a large language model find that and say, oh, this article says Mount Everest is this high, and that article says Mount Everest is that high. Hey, let’s, let’s dig in. And I think that kind of stuff, you know, using technology to help the community improve Wikipedia seems like a perfectly sensible thing to do, but I think it’s gonna be a long time before you don’t need like, real human oversight because these AIs aren’t that good yet.

 

ISAACSON: I’ve been using Grokopedia for the past few days, and I was somewhat surprised at how it does try to do balance both things, even on controversial things. And I even asked Grokopedia, tell me about Jimmy Wales and tell me about Wikipedia. And what it says is “conservative analysts decry Wikipedia’s over-reliance on media ecosystems for sourcing leading to the underrepresentation of right-leaning data, while progressive defenders praise Wikipedia’s inclusivity to marginalized narratives as aligning with evidence-based consensus.” In some ways, that reads like a Wikipedia entry, giving both sides of an argument. 

 

WALES: Yeah, yeah. No, it does, it does. That’s a good summary of the debate for sure.

 

ISAACSON: But it does go on to say that in some ways you, you’re – “even though it has unprecedented scale, it nonetheless has a bias towards the mainstream media,” Wikipedia does. Is that a fair criticism? ’cause I think Larry Sanger, who was at there at the founding of the Wikipedia with you, makes that criticism as well.

 

WALES: Well, it, it’s, it’s fair, but is it valid as a criticism? I mean, it’s, it’s of course, true that we don’t treat fringe sources, random blogs, tweets from crackpots as being the equivalent of fact-based researched information. And I, I’m unapologetic about that. So, you know, if you say, wow, you know, Wikipedia treats, you know, academic medical journals as being more important than random social media influencers, I’m just gonna plead guilty unapologetically and say, those are not the same thing and they shouldn’t be the same thing. Like, getting facts right is really important and you can’t treat fringe sources as if they’re just the equal of mainstream sources. There’s a reason they’re mainstream, which is that they’re tried, true, tested, et cetera. That doesn’t mean the mainstream’s always right, of course. Right. We all know of instances where, you know, the world thought one thing and then over time evidence emerged and, and the understanding of the world changed. That’s also part of the process. But I don’t think you can say we we’re gonna do away with all standards and treat every random utterance on the internet as completely equal. That’s crazy.

 

ISAACSON: You talk about neutrality and sometimes these days it’s very hard. One of the criticisms, I think US Congress even did it, is the extent to which you cover Gaza and Israel and stuff. And  I’d like to read you, what is the beginning of what’s called “Gaza genocide” in Wikipedia. I just read it last night. It says, “the Gaza genocide is the ongoing, intentional and systematic destruction of the Palestinian people in the Gaza Strip carried out by Israel” and it doesn’t say this is Israel’s side of whatever. Is there some sense that these type of things can happen in Wikipedia where it does have a particular bias?

 

WALES: I mean, they can, I think that’s one of the worst Wikipedia entries I’ve seen in a very long time. I just looked at it myself and I was shocked to see it. It needs to change. I think it’s just terrible. It doesn’t live up to our standards of neutrality, and I think it’s very problematic. And it’s definitely something that would not reflect on the way Wikipedia should work and we need to fix it. I, that’s all I could say to that. I mean, I wish I could be a, a PR defender and say, no, no, of course Wikipedia’s always perfect, but that wouldn’t build people’s trust. I think people will have more trust in Wikipedia if I acknowledge, like, you know what, we don’t always get it right. And actually what we need are kind and thoughtful people to come in and join the community and help us fix anything that’s not really fair

 

ISAACSON: Because of the entry on Gaza and other things, the Congress has said that they want a subpoena to say who wrote those things. Your people are quasi anonymous. They write under pseudonyms most of the time. Do you think it’s okay to give up the names of the people who write for Wikipedia?

 

WALES: I think it’s deeply inappropriate for the US government under the First Amendment to start attacking people for being biased. Like that is, there’s no, like, it’s completely not in the American tradition to do that sort of thing. It’s very McCarthy era-esque. And so yeah, clearly we are gonna fight against that sort of thing. Like that’s a ridiculous thing for Congress to do.

 

ISAACSON: Let me read you a quote that you said, which is, “the sunny pro-social view of human nature that inspired Wikipedia may be out of fashion” “But I will insist that it’s correct.” Unpack that for me, please.

 

WALES: Yeah, you know, I mean, the thing is, we all go around and we meet other people. And as, as we well know in our normal lives, we think most people are basically decent. You know, some people are   annoying and all of that, but the number of truly malicious people in the world is actually very small, and people love to come together and do positive things. We see all kinds of examples of this you know, charity works or anytime there’s a natural disaster, people come together and they help each other and they do that in a really positive spirit. And I think we are at risk of forgetting that when we are thinking about human nature through the lens of looking at social media platforms that prioritize and promote the most toxic people. And so suddenly you get this idea like, wow, the world is full of angry, upset, like horrible people. When you look around you and you’re like, no, actually most people are really nice. And actually that’s kind of weird. Like, those people become prominent on social media because of bad algorithms that are like really prioritizing the wrong things.

 

ISAACSON: One of your rules is: assume good faith. Does that work?

 

WALES: It does, it does. You know, it turns out a there’s a, a great story in the book about Kaylana, that’s her, that’s her username, but she started editing Wikipedia as a teenager and has become really respected Wikipedian, but she started her first ever edit to Wikipedia was vandalism because she couldn’t, you know, believe like you could really do this. And then somebody came to her and said, Hey, you know, like we’re talking –

 

ISAACSON: In other words, she vandalized one of the pages. She wasn’t writing about the concept of vandalism.

 

WALES: Right. Yeah, yeah. She vandalized one, she wrote some nonsense, you know, and somebody stopped her and said, Hey, like, we’re not doing that. She felt so bad, like, she’s like, oh, like these nice people are being nice to me, and they assume I was just doing a test or whatever, that she then turned into this fantastic Wikipedian. She’s like, oh, that’s amazing. Like, and so, you know, and we know this, like, you meet somebody new for the first time and you don’t instantly think, okay, what’s this horrible person gonna do to me? In fact, if that is how you feel about meeting new people, like that’s a problem in your life. Like, you need to work on that. Generally, like yeah, approach people open heart and say, oh, great, yeah, nice to meet you. Like let’s get to know each other. That assuming good faith of people that even if they’re doing something you don’t agree with, that probably their heart’s in the right place somehow really pays off almost always. Like sometimes no, like sometimes people are just horrible. That’s, that’s true too. But broadly just that initial assumption of like, Hey, let’s do something good together, it really does work.

 

ISAACSON: I once early on had an experience with Wikipedia, and it was on the Einstein entry, and I was writing a book about Einstein and it kept saying that Einstein had been given a visa by the King Zog of somewhere to – and it was totally wrong. And so I went in and I corrected it, and yet the people who were sort of partisans of that kept putting it back in. And finally I cited the things I gave the sources and did, and it stayed. And I said to myself, boy, that proves you need experts doing it. I was bragging, thinking of myself as an expert instead of the wisdom of crowds. And then it occurred to me, I’m just part of the crowd and I was one of the many people contributing. Is that sort of the magic of how that works?

 

WALES: Yeah, and it was, it’s also a good illustration of how, like, good quality, reliable sources are crucial because if it’s just random people’s opinions up against each other, how do you decide? But if you’ve got, like, a source, a historian, a documented fact, great, brilliant. That actually moves the conversation forward in a very good way.

 

ISAACSON: Jimmy Wales, thank you so much for joining us. Appreciate it. 

WALES: Thank you. Great. Fantastic.