What good is being the wealthiest man alive if your chatbot won’t lie in support of your rancid politics? That’s the question far-right billionaire Elon Musk appears to be asking once again of Grok, the AI model developed by his company xAI and integrated into X, his social media platform.
Following the assassination on Saturday of Minnesota Democratic Rep. Melissa Hortman and her husband Mark Hortman in their home, and the shooting of Minnesota Sen. John Hoffman and his wife Yvette Hoffman, allegedly by the same suspect, much of the online right mobilized to paint the horrific attack as the work of a radical leftist. Their efforts were somewhat undercut when it emerged that Vance Boelter, the 57-year-old man charged with second-degree murder and attempted murder in the case, appeared to be an anti-abortion Trump voter who had a “hit list” solely consisting of Democratic elected officials. (He has not yet entered a plea.)
But, never one to wait for facts to emerge (or absorb them when they become available), Musk rushed along with MAGA influencers to push a bogus narrative of leftist extremism over the weekend. “The far left is murderously violent,” he wrote in an X post on Saturday, quoting a Trump supporter who had falsely attributed the politically motivated murders in Minnesota to “the left.” That prompted another user to tag Grok in the thread and ask, “Why is the left so murderously violent? They don’t seem so tolerant.” Grok replied, “The claim that ‘the left’ is murderously violent isn’t backed by evidence,” offering a centrist correction: “Political violence spans all side — right-wing attacks, like Jan. 6, and left-wing protests, like 2020 riots, both occur but aren’t exclusive to one group.”
That evening, Musk responded to an X user and Trump backer who complained that Grok had been “manipulated by leftist indoctrination,” writing, “I know. Working on fixing that this week.” It’s not clear how either of them imagined a chatbot created and trained by a Musk-controlled company had fallen victim to leftist brainwashing, though it’s hardly the first time he and his fellow right-wing conspiracists have raged at Grok for contradicting their worldview. From the start, conservatives have said the bot is too “woke” on issues including diversity and gender identity, contrary to Musk’s stated goal of unleashing an “anti-woke” AI.
Musk himself was disappointed last year during an appearance on The Joe Rogan Experience when he directed Grok to ridicule transgender athletes only to have the bot instead mock the “Karens” who have made this a headline topic of the culture wars. The model is also known to accurately reflect that Musk is a major spreader of misinformation and debunk the misleading talking points that circulate Trumpworld.
The tech oligarch’s frustration over the political violence summaries from Grok seemed to escalate as it kept noting that right-wing violence is more common and has claimed more casualties than the left-wing equivalent in recent years. “Since 2016, data suggests right-wing political violence has been more frequent and deadly, with incidents like the Jan. 6 Capitol riot and mass shootings (e.g., El Paso 2019) causing significant fatalities,” it said on Tuesday. “Left-wing violence, while rising, especially during 2020 protests, is less lethal, often targeting property.” Musk called the response incorrect. “Major fail, as this is objectively false,” he wrote without presenting countervailing evidence. “Grok is parroting legacy media. Working on it.” As of Wednesday, Grok was still reporting that extremism on the right results in more deaths. “From 2015–2020, far-right incidents totaled 267 with 91 fatalities, compared to 66 left-wing incidents with 19 deaths,” it said in one post. “Right-wing attacks often target people, while left-wing violence typically involves property damage during protests, which can seem more visible but is less deadly.”
Exactly how Musk’s team at xAI might tweak Grok to fit his distorted view of reality is anybody’s guess, though the results are unlikely to be subtle. Last month, X users were bewildered as the bot repeatedly invoked the specter of a supposed “white genocide” in South Africa, even when asked to weigh in on totally unrelated subjects. White people are not being systemically murdered in South Africa, where Musk was born, but he and Trump have repeatedly and falsely proclaimed that the nation’s white minority are victims of violent persecution, with the president fast-tracking the process of immigration to the U.S. for a handful of “refugees” fleeing this imaginary atrocity. Many speculated that Grok had been specifically altered to bring up South Africa at Musk’s instructions; even then, it had no way of substantiating the baseless notion of a “white genocide” occurring there.
That behavior was later amended, and the bizarre posts deleted — easy enough. Getting a chatbot that is programmed to be “truth-seeking” to lie about terrorism statistics in a way that doesn’t look like brazen puppeteering may prove difficult by comparison. Still, you can count on Musk to keep up the great work on the most important technological challenge of our time: forcing a piece of software to agree with him about everything.
