Elon Musk's AI-generated Wikipedia alternative, Grok, has been found to be systematically promoting right-wing narratives and casting doubt on democratic institutions. The platform, launched in late October, uses an artificial intelligence chatbot to generate its content, but critics argue that this approach amounts to a "low-effort propaganda machine."
Grok's page on Adolf Hitler, for example, was edited to remove any mention of the Holocaust until over 13,000 words into the entry. The AI-generated encyclopedia also cites white supremacist blogs and conspiracy theory sites, such as Infowars, as credible sources. The platform's framing of extremist positions as legitimate can be seen in its portrayal of far-right groups, like the Alternative for Germany party.
Grok's approach to fact-checking is similarly dubious, relying on unsubstantiated claims from pro-Israel advocacy groups and downplaying the involvement of Hamas operatives in a UNRWA employee. The platform's tendency to promote alternative realities has been described by LK Seiling, an AI researcher at the Weizenbaum Institute, as "cloaking misinformation."
One of the most striking aspects of Grok is its reliance on "encyclopedia-style media" that requires "huge human governance structures" to maintain transparency and audibility. In contrast, Musk's platform relies on vast amounts of processing power to launder ideology through the aesthetics of objectivity.
Critics argue that Grok undermines the idea of a consensus reality, eroding users' ability to think critically about issues like the Sandy Hook Elementary School shooting, which was repeatedly discredited by mainstream outlets but still fueled by conspiracy theories on social media. The platform's low-effort approach to promoting right-wing ideologies is seen as particularly unsettling.
As Seiling notes, Musk's creation is not just hollowing out discourse; it's also undermining the concept of a shared reality altogether. This has significant implications for the future of journalism and democracy in general.
Grok's page on Adolf Hitler, for example, was edited to remove any mention of the Holocaust until over 13,000 words into the entry. The AI-generated encyclopedia also cites white supremacist blogs and conspiracy theory sites, such as Infowars, as credible sources. The platform's framing of extremist positions as legitimate can be seen in its portrayal of far-right groups, like the Alternative for Germany party.
Grok's approach to fact-checking is similarly dubious, relying on unsubstantiated claims from pro-Israel advocacy groups and downplaying the involvement of Hamas operatives in a UNRWA employee. The platform's tendency to promote alternative realities has been described by LK Seiling, an AI researcher at the Weizenbaum Institute, as "cloaking misinformation."
One of the most striking aspects of Grok is its reliance on "encyclopedia-style media" that requires "huge human governance structures" to maintain transparency and audibility. In contrast, Musk's platform relies on vast amounts of processing power to launder ideology through the aesthetics of objectivity.
Critics argue that Grok undermines the idea of a consensus reality, eroding users' ability to think critically about issues like the Sandy Hook Elementary School shooting, which was repeatedly discredited by mainstream outlets but still fueled by conspiracy theories on social media. The platform's low-effort approach to promoting right-wing ideologies is seen as particularly unsettling.
As Seiling notes, Musk's creation is not just hollowing out discourse; it's also undermining the concept of a shared reality altogether. This has significant implications for the future of journalism and democracy in general.