Wikipedia is a treasured on-line useful resource that, regardless of huge modifications throughout the net, has managed to remained actually nice to at the present time. I, alongside tens of millions of different customers, go to the positioning day by day to be taught one thing new or double-check present data. In an age of continuous AI slop, Wikipedia is one thing of an antidote.
When you take a look at Wikipedia and suppose “that is alright, however an AI model could be loads higher,” you may simply be Elon Musk. Musk’s AI firm, xAI, simply launched Grokipedia—sure, that basically is its title—a web-based encyclopedia that carefully resembles Wikipedia in title and surface-level look. However beneath the hood, the 2 might hardly be any extra completely different. Although it is early days for the brand new “encyclopedia,” I might say it isn’t value utilizing, no less than not for something actual.
The Grokipedia expertise
Whenever you load up the Grokipedia web site, it appears pretty commonplace. You see the Grokipedia title, alongside the model quantity (v0.1, on the time of writing), alongside a search bar and an “Articles Obtainable” counter (885,279). Looking for an article too is fundamental: You kind in a question, and a listing of obtainable articles seems so that you can choose from. When you pull up an article, it appears like Wikipedia, solely extraordinarily fundamental: There are not any photographs, solely textual content, although you need to use the sidebar to leap between sections of the article. You may additionally discover sources, famous by numbers, which correspond to the References portion on the backside of every article.
The important thing distinction between Grokipedia and a easy model of Wikipedia, nonetheless, is that these articles are usually not written and edited by actual individuals. As an alternative, every article is generated and “fact-checked” by Grok, xAI’s giant language mannequin (LLM). LLMs are in a position to generate giant quantities of textual content briefly durations of time, and embrace sources for the place they pull their data, which could make the pitch for Grokipedia sound nice to some. Nevertheless, LLMs additionally tend to hallucinate, or, in different phrases, make issues up. Typically, the sources the AI is pulling from are unreliable or facetious; different instances, the AI takes it upon itself to “lie,” and generate textual content that merely is not true. In each circumstances, the data can’t be trusted, particularly not at face worth, which is why it is troubling to see a lot of the expertise is fully powered by Grok, with out human intervention.
Grokipedia vs. Wikipedia
Musk is pitching Grokipedia as a “huge enchancment” over Wikipedia, which he has criticized for pushing propaganda, notably in the direction of left-leaning concepts and politics. It is ironic, then, that a few of these Grokipedia entries are themselves pulling from Wikipedia. As The Verge’s Jay Peters highlights, articles like MacBook Air observe the next on the backside: “The content material is tailored from Wikipedia, licensed beneath Inventive Commons Attribution-ShareAlike 4.0 License.” What’s extra, Peters discovered that some Grokipedia articles, akin to PlayStation 5 and the Lincoln Mark VIII, are virtually one-to-one copies of the corresponding articles on Wikipedia.
When you’ve adopted Musk’s politics and political actions lately, it will not shock you to be taught he falls on the right-wing aspect of the political spectrum. That may give pause to anybody who considers utilizing Grokipedia as an unbiased supply of data, particularly as Musk has repeatedly retooled Grok to generate responses extra favorable to right-wing opinions. Critics like Musk declare Wikipedia is biased in the direction of the left, however Grokipedia is fully produced by an AI mannequin with an abject bias.
You may see that you’ve very completely different experiences when studying sure subjects throughout Wikipedia and Grokipedia. Wikipedia’s Tylenol article, for instance, reads the next:
In 2025, Donald Trump made a number of statements a couple of controversial and unproven connection between autism and Tylenol. These statements, concerning the connection between Tylenol throughout being pregnant and autism, are based mostly on unreliable sources with out scientific proof.
Evaluate that to Grokipedia, which devotes three paragraphs to the topic, the primary of which begins:
A number of observational research and meta-analyses have recognized associations between prenatal publicity to acetaminophen (the energetic ingredient in Tylenol) and elevated dangers of neurodevelopmental problems (NDDs) in offspring, together with attention-deficit/hyperactivity dysfunction (ADHD) and autism spectrum dysfunction (ASD).
That mentioned, the second paragraph highlights a few of the points with these research, whereas the third highlights that sure companies recommend the “advantages outweigh unproven dangers.”
What do you suppose to date?
Equally, as noticed by WIRED, Grokipedia’s article, Transgender, highlights the idea that social media might have acted as a “contagion” to the rise in transgender identification. Not solely is {that a} frequent right-wing assertion, that individual phrase might have been plucked from a put up from a right-wing X account. Wikipedia’s article, as you may count on, doesn’t entertain the declare in any respect.
Grokipedia can also be favorable to unproven, controversial, or flat-out absurd claims. As Rolling Stone highlights, it refers to “Pizzagate,” a conspiracy concept that result in a real-life taking pictures, as “allegations,” a “speculation,” and a “narrative.” Grokipedia offers credence to “Nice Substitute,” a racist concept floated by white supremacists.
Must you use Grokipedia?
This is the brief reply: no. The difficulty I’ve with Grokipedia is two-fold: First, no encyclopedia goes to be dependable when it’s virtually fully created by AI fashions. Positive, a few of the data could also be correct, and it is nice you’ll be able to see the sources the bot is utilizing, however when the danger of hallucination is baked into the expertise with no approach round it, selecting to keep away from human intervention en masse all however ensures inaccuracies will plague a lot of Grokipedia’s data base.
As if that wasn’t sufficient, this Grokipedia is constructed on an LLM that Musk is overtly tinkering with to generate outcomes that extra carefully align together with his worldview, and the worldview of 1 specific political ideology. Hallucination and bias—simply the components you want for an encyclopedia.
The factor about Wikipedia is it is written and edited by people. These people can maintain different human writers accountable, including new data when it turns into obtainable and correcting errors after they encounter them. Maybe it is irritating to learn that your favourite well being and human providers secretary “promoted vaccine misinformation and public-health conspiracy theories,” however that is the target, scientific actuality. Eradicating these goal descriptions, and reframing the dialogue in a approach that matches a warped worldview would not make Grokipedia higher than Wikipedia—it makes it ineffective.
