“The goal here is to create an open-source, comprehensive collection of all knowledge,” said Elon Musk on X, as his xAI company rolled out its first version of AI-powered online encyclopaedia, Grokipedia.
Having already set out to revolutionise electric cars, explore space, upend social media, and roll back the state, Musk’s latest venture is “something altogether more fundamental: a new version of the truth”, said Jemima Kelly in the Financial Times.
‘A new version of the truth’
Named after X’s built-in AI factchecker, Grok, the origins of Grokipedia date back to the end of last year, when Musk told followers to “stop donating to Wokepedia”. Accusing Wikipedia of spending too much money on diversity, equity, and inclusion, he branded the online encyclopaedia “an extension of legacy media propaganda”.
Things ramped up in late September, when Donald Trump’s AI czar David Sacks posted on X that Wikipedia was “hopelessly biased”, saying “an army of left-wing activists maintain the bios and fight reasonable corrections.”
While there may be some commercial motivation at play, Filippo Trevisan, an associate professor of public communication at American University in Washington DC, told DW, the true impetus behind the project is ideological. Grokipedia “responds to those criticisms of Wikipedia from so many figures within the American conservative and the right-leaning world”. This is Musk’s bid to “present AI as a solution to the bias problem”.
“There is a growing belief that algorithmic aggregation is more trustworthy than human-to-human insight,” David Larsson Heidenblad, deputy director of the Lund Centre for the History of Knowledge in Sweden, told The Guardian. The “Silicon Valley mindset” focuses on learning through trial and error, in contrast to the traditional academic process of “building trust over time and scholarship over long periods”.
How different is it from Wikipedia?
Given the deep hostility towards Wikipedia, it is odd that Grokipedia appears to use the site as its “primary source”, said Vox. On other topics, however, it “injects some far-right politics and conspiracy theories into certain topics before presenting the information as fact”. On launch there was, for example, no article on “apartheid”, but a defence of “white genocide theory” – “one of Musk’s ideological obsessions and the centre of many unhinged Grok rants earlier this year”.
While many of the pages appear “fairly similar” to Wikipedia “in terms of tone and content”, said Wired, a “number of notable Grokipedia entries denounced the mainstream media, highlighted conservative viewpoints, and sometimes perpetuated historical inaccuracies”. In one instance, an entry made the unsubtantiated claim that “the proliferation of porn exacerbated the HIV/Aids epidemic in the 1980s”.
“The main distinction between the two comes in how information is checked and processed,” said DW. “Wikipedia relies on collaborative community editing”, with processes in place to identify and correct errors. Grokipedia has no human editorial involvement and appears to “lack such oversight”, Roxana Radu, associate professor of Digital Technologies and Public Policy at Oxford University, told the news site.
“Instead of setting up a serious challenger to Wikipedia, Musk has scored a major own goal,” said Kelly in the FT. Grokipedia demonstrates that, “while humans might be highly imperfect, biased and tribal beings, they are still better than AI at getting to the truth”.
AI-powered online encyclopaedia seeks to tell a ‘new version of the truth’