Elon Musk has lambasted Wikipedia, calling it woke, supporting accusations of it having a left-wing bias, and arguing it’s an “extension of mainstream media propaganda.” So his AI firm, xAI, developed a rival referred to as Grokipedia, which paradoxically sufficient cites the web encyclopedia on the finish of a lot of its articles.
Grokipedia.com got here on-line on Monday, and options 885,279 articles on the positioning as of Tuesday. The net encyclopedia is Musk’s reply to what he’s repeatedly referred to as “Wokipedia,” the favored free on-line supply with over 7 million articles in English alone as of Tuesday. Wikipedia articles are maintained by volunteer writers and editors, whereas xAI’s service makes use of AI to generate articles by pulling from totally different sources throughout the web, one being Wikipedia.
The launch of Grokipedia marks its preliminary model 0.1, however a later model 1.0 “will be 10X better,” Musk stated in an X publish on Monday.
“But even at 0.1 it’s better than Wikipedia imo [in my opinion],” Musk wrote.
On the finish of articles on Grokipedia, the reader will get a discover that the content material was tailored by Wikipedia beneath a Inventive Commons Attribution-ShareAlike 4.0 License, which is a content material license that permits anybody to repeat, distribute, adapt and construct upon inventive works together with textual content, pictures and music—even for business functions—so long as they supply correct attribution to the unique creator. The “Share Alike” portion of the license means the creator that makes a brand new model of content material beneath this license is agreeing to permit others to do the identical to their content material.
xAI’s massive language mannequin (LLM), Grok, which powers the brand new web site, seems at a supply of knowledge like a Wikipedia web page and determines “what is true, partially true or false, or missing,” Musk stated throughout the All-In Summit in September.
In its solutions to consumer prompts, Grok will “rewrite the page to remove the falsehoods, correct the half truths and add the missing context,” he added.
A 2024 report by the Manhattan Institute, a conservative-leaning public coverage suppose tank in New York Metropolis, discovered tendencies that “constitute suggestive evidence of political bias in Wikipedia articles,” together with evaluation discovering that names of distinguished left-leaning U.S. politicians tended for use “with more positive sentiment than their right-leaning counterparts.”
“We want to acknowledge Wikipedia’s significant and valuable role as a public resource,” the research’s authors wrote. “We hope this work inspires efforts to uphold and strengthen Wikipedia’s principles of neutrality and impartiality.”
Fortune’s request for remark from xAI prompted a seemingly computerized response that learn: “Legacy Media lies.”
