Happy Birthday, Wikipedia: We need you now more than ever
This week the best website in the history of the internet celebrates its 25th birthday. What URL could I possibly mean besides Wikipedia.org — bastion of what the internet once was, and what it still could be. It stands in stark contrast to nearly every other online platform that has been ens**ttified by profit-hungry algorithms, AI slop and racist bots. The World Wide Web can feel like a pretty dark place lately, but on Wikipedia — which exists in 340 languages, hosts 7.1 million articles in English, and is consistently among the top 10 most visited websites — it feels a lot different, much brighter and more free.
Its authors, as you probably know, are countless and largely anonymous “Wikipedians,” volunteer editors who nitpick citations and uphold its editorial standards so you and I can enjoy entries about weird mushrooms, the cast of “The Pitt” or the latest political crisis in crowd-sourced and generally fair-minded terms. And Wikipedia is still improving steadily, edit by edit, which is remarkable given how close it has remained, over all those years, to its original principles and site design. It’s not just a website for settling an argument — though it’s really good for that! — it’s a place to satisfy curiosity, brush up on basics or take a wild trip down an information rabbit hole and learn things you didn’t know you needed to know.
That’s not to say Wikipedia is perfect. It has its well-documented gender and cultural biases, all of them spelled out in detail on its own pages, indicating how far it still has to go. Nonetheless, for a project as massive and long-lived as Wikipedia, it’s surprising it hasn’t been strung upside down, its pockets shaken out and left to bleed. That’s pretty much what happened to MTV, Vice Magazine, corporate DEI statements, Twitter and quite a few other big ideas that once seemed unstoppable but have now faded away ignominiously, largely thanks to incompetence, greed and animosity.
Somehow, Wikipedia has done a fair job of dodging those most awful elements of human nature, yet its future may be less rosy than it seems. Stalking the website, knives out like a bromidic comic book clone, is Grokipedia, the brainchild of Elon Musk, the world’s richest man and biggest toddler.
Launched in October 2025, Grokipedia is one of the most impressive propaganda campaigns ever undertaken in human history. Boasting six million articles and counting, it completely dwarves similar endeavors like Conservapedia (a conservative Wikipedia fork that boosts “Young Earth Creationism,” the hypothesis that the universe is only a few thousand years old) or Ruwiki, a “Putin-friendly” counterpart to Wikipedia, the latter being generally unavailable in Russia.
Start your day with essential news from Salon.
Sign up for our free morning newsletter, Crash Course.
But all those alterna-wikis still rely on human hands to type their falsehoods. The articles on Grokipedia are unique, in an especially bad way: They are composed by the infamous large language model Grok, a product of Musk’s company xAI. (Most people would call Grok an “AI,” but I won’t; artificial intelligence, in its current usage, is largely a meaningless marketing term.) Though the vast majority of Grokipedia articles have been copy-pasted straight Wikipedia, the similarities end there. There are no images and no attempt at engaging design, and the prose reads exactly as if it were written by a robot. (Because it was.) The Grokipedia homepage is, um, minimalist, meaning bland and executed without imagination.
More important by far, while users can propose edits on Grokipedia, they can’t make them. This has already led to considerable conflict over what Grok considers a reputable source or a conspiracy theory. In countless instances, Grokipedia has promoted pseudoscientific hypotheses, including the ones about vaccines causing autism and HIV not causing AIDS, as well as transphobic and racist views on everything from the Holocaust to George Floyd’s murder. It employs invented or low-quality sources, sometimes citing its own Grok chats. It’s the Ouroboros equivalent of “trust me, bro.”
Let’s give credit where credit is due: Musk, or whoever came up with the name Grok, is a marketing genius. That’s exactly the sound my dog makes when he vomits, which is a useful metaphor for what Grok does. It regurgitates information that sometimes resembles the truth spews it out on the world. Grok is sometimes functional as a tool, but given its sporadic tendencies to praise Hitler and generate sexualized images of children, it’s pretty dangerous. If my car ran OK most of the time but occasionally shot flames from the exhaust, I wouldn’t consider it a reliable machine.
Grokipedia isn’t about seeking the truth — it’s about muddying the waters of what you can trust online.
Here’s a fun fact: Wikipedia was originally part of a project called Nupedia, and thank God we didn’t get stuck with that name. The term “wiki” comes from the Hawaiian word for “quick,” with Wikipedia taking its name from the Wiki Wiki shuttles at Honolulu airport. (This wouldn’t be an article about Wikipedia without tons of tangents and open tabs.) As for Grok, it was originally going to be named TruthGPT, perhaps in a nod to the Orwellian doublespeak of Truth Social, Donald Trump’s answer to Twitter. Of course, “grok” — a term coined by Robert E. Heinlein in “Stranger in a Strange Land” — means to “understand something intuitively or by empathy.” But Grokipedia doesn’t operate like that at all.
In so many ways, Grokipedia is a reflection of its mastermind. It is petty, arrogant and misleading, expressing ill-informed and overtly bigoted worldviews. It doesn’t invite engagement, but serves itself up as the final word. It has sacrificed integrity, humility and accuracy for speed and scale. It’s just more “move fast and break things” Silicon Valley ideology, thinly covering for carelessness and zero accountability. In other words, all the same attitudes that Musk perpetuates in his business decisions, broken puns and desperate pleas for attention.
Let’s not forget why Musk “invented” Grokipedia in the first place. Back when the South African-born billionaire was still posturing as a nerdy genius and not a fascist wannabe, Musk held Wikipedia in high esteem. But as his behavior and opinions leaned further right, he became increasingly irritated with pages such as “Views of Elon Musk.” He once offered to give the Wikimedia Foundation, which operates the site, “a billion dollars if they change their name to Dickipedia.” Then, after Musk helped Trump win the 2024 election, came his semi-deniable Nazi-salute incident. Musk was greatly displeased with the resulting edits to his Wiki biography, but perhaps he should’ve considered the repercussions in advance. After that, Musk’s war against “Wokeipedia,” as he called it, became a full-frontal campaign for a man with more money than God and skin thinner than graphene, the thinnest two-dimensional material in the world. (Yes, I looked that up on Wikipedia.)
Musk quickly found out that he couldn’t influence Wikipedia quite as easily as he could buy Twitter and shred its credibility. So he created his own version, which has very little to do with seeking “the truth, the whole truth and nothing but the truth,” as Musk put it, and much more to do with muddying the waters about who and what you can trust online.
It’s not easy to create an encyclopedia, even if you have a bot doing all the work for you. Bill Gates found that out as much with the development of Microsoft Encarta, the encyclopedia project spearheaded by his mega-corporation in the 1990s. (Remember when encyclopedias used to exist on CD-ROMs? Must be a generational thing.) When criticism arose about factual discrepancies in different languages of Encarta, Gates penned an op-ed in the Sunday Times in 1997 explaining why some facts differed:
But when international versions of Encarta eventually go up on the Internet, our policy of presenting “local, educated reality” will be called into question. Some readers will get upset about content that may fly in the face of their reality.
A Korean reader may gain access to Japanese Encarta and note that the East Sea is called the Sea of Japan. Some French readers may be offended by how much media is devoted to the English article on the Battle of Waterloo, in which Napoleon was finally defeated.
In the long run, exposing people to worldwide perspectives should be healthy.
Americans will benefit from a better understanding of Asian or European views of important cultural and scientific events, and vice versa.
One of the challenges of presenting history is its inability to be captured by a single interpretation. The goal, therefore, must be to find neutral ground and present opposing points of view where appropriate.
Truth must not be a victim of this process.
Such an attitude is completely nonexistent at Grokipedia. There is no indication that knowledge is sticky or complicated. Musk’s goal, in Grokipedia and basically everywhere else, is to pose as a definitive source and silence voices he doesn’t like. As for its practical utility, Musk would argue that Grokipedia is still in its early stages and will eventually become so good that we’ll forget all about its shortcomings or errors. It’s not even a year old! Only version 0.2! In 25 years, he may contend, Wikipedia will be a distant memory, an answer to a trivia question just as Nupedia is now.
Elon Musk’s goal, in Grokipedia and everywhere else, is basicaly to pose as a definitive source and silence voices he doesn’t like.
It’s not spiteful or cynical to suggest that seems unlikely. Musk has simply missed the point of Wikipedia, just as he has missed the point of so much else. This is reflected in the fact that Grokipedia’s traffic has cratered since launch. It has millions of articles, but hardly anyone is reading them.
What has made Wikipedia a mainstay of human culture is that humans authored it. It has something of a bureaucracy problem, and some of its infamous edit wars have waged for decades. But that sort of friction is precisely what has made Wikipedia a trusted source. Wikipedia has never posed itself as the final word. It is best at being surface-level information but efficiently organized and well-sourced with citations that are a joy to follow. Going down rabbit holes or adding a link where it says “citation needed” is the whole point. It will never be perfectly accurate, because human knowledge will never be right about everything, but at least we can trust that actual humans are doing their best to arrive at the most likely conclusion. We can’t say the same for a machine operated by a narcissistic billionaire who is troublingly eager to prove he’s smarter than the rest of us.
Read more
about artificial intelligence