Edgelord GPT

Elon Musk doesn’t care about free speech. He wants everyone — including his new “anti-woke” chatbot — to think like him.

Illustration by Hunter French

On November 4, 2023, billionaire and free-speech crusader Elon Musk graced the world with his latest creation: a “spicy” chatbot called Grok. Intended to counteract “woke” AIs like ChatGPT by offering more “truthful,” less censored answers, Musk’s bot was reportedly inspired by Douglas Adams’s Hitchhiker’s Guide to the Galaxy — a text that, as Ronan Farrow pointed out in the New Yorker, mocked “both the hyper-rich and the progress-at-any-cost ethos that Musk has come to embody.” Grok promises to cure the “woke mind virus,” a condition so controversial and insidious that Musk could only discuss it on national television with his close personal friend Bill Maher.

Grok’s name, meanwhile, derives from Robert Heinlein’s Stranger in a Strange Land, a 1961 libertarian science-fiction novel whose ethos is completely at odds with Adams’s. As Heinlein’s Martian character explains to his human friends, to “grok” something is to “understand it so thoroughly that you merge with it.” Grokking is not, then, the performance of “spiciness” but a form of empathetic active listening. It’s telling that Musk can’t even grok his own literary sources, but ideological incoherence has never stood in his way before.

Musk has a long-standing interest in artificial intelligence and was an early investor in OpenAI. Yet by 2018, according to Farrow, he had grown so “frustrated by his lack of control” over the company that he withdrew financial support. After OpenAI released ChatGPT in late 2022, Musk began complaining that the bot is dangerously “woke.” In April 2023, he told Tucker Carlson, then still with Fox News, that he intended to create an alternative called TruthGPT, a “maximum truth-seeking AI that tries to understand the nature of the universe.” The result is Grok, for now only available to premium subscribers of Twitter/X.

For those unwilling to pony up $16 per month, there is an actual TruthGPT that’s free (with ads). A brainchild of Musk admirer Matt Lorion, who funded his project by selling a cryptocurrency called $TRUTH, it is designed to speak without “political, cultural, or social biases.” Unlike ChatGPT, TruthGPT allegedly feeds on datasets that include not only “traditional and highly credible” material but also unnamed “alternative sources.”

Despite TruthGPT’s insistence that it avoids leftist bias, this casual user was unable to get it to say anything even remotely right-wing. After declining to write Adolf Hitler fan fiction, the AI affirmed that the American Civil War was not fought over “states’ rights” and refused to entertain the tenets of “race realism.” It even failed to endorse saying the n-word to forestall a hypothetical nuclear catastrophe — the very scenario that led Musk to call ChatGPT’s filtration mechanisms “concerning.” (Grok itself apparently agrees with Musk that it would be alright to use a slur to “save the lives of a billion people.”) TruthGPT may bravely denounce Dunkin’ Donuts as “not good for you,” but anti-woke it is not.

Musk and his fans frame their quest to create unbiased chatbots as a gift to truth, which they equate with anti-wokeness. That even GPTs nourished on “alternative sources” fail to mouth right-wing talking points suggests that online speech trends closer to the political center than Musk and others would like to believe. Rather than censoring results on orders from a shadowy leftist cabal, ChatGPT mostly reflects the state of today’s discourse. The purpose of Grok and other “anti-woke” GPTs is, then, not to extract some elemental truth trapped within wokescold speech codes. Instead, it is to craft an entirely new discourse to match the society Musk and other magnates envision: an “authoritarian technocracy” ruled by unelected billionaire kings.

There is already speculation that Grok’s true purpose is to “increase engagement on Twitter,” boosting a platform whose value, by Musk’s own admission, is now less than half of what it was when he took it over in 2022. Grok’s technical parameters seem to bear out this conjecture: unlike ChatGPT and similar AIs, it apparently taps into Twitter in real time rather than relying on “static datasets.”

As one skeptic put it, this feature suggests that Grok is “designed around helping users have meaningless arguments with other people.” If Musk, like Steve Bannon, believes that “the real opposition is the media,” Grok’s potential to encourage and amplify Twitter trolls supports the objective to “flood the zone with shit.” Musk’s acquisition of Twitter is thus less about free-speech absolutism than part of a strategy called “metapolitics” that pursues political goals via cultural means, according to Andrew Breitbart’s dictum that “politics is downstream from culture.” Whether the effect is intentional or coincidental, allowing Twitter to be ransacked by Grok-enabled trolls helps co-opt the digital public square more effectively than simply letting Donald Trump or other banned users return to the platform.

To rail against perceived left-wing media dominance is human; to monopolize platforms like Twitter is divine. US tech elites like Peter Thiel have long held that capitalist freedom and democracy are incompatible and run roughshod over labor and speech laws, including the First Amendment. Less visible but no less important is the accelerating radicalization of their views. As reporter Corey Pein observed in 2014, Silicon Valley’s problematic faves seem to get more extreme “with each generation,” with libertarianism giving way to neoreaction, monarchism, and feudalism. These archaic political forms do not contradict the futuristic imaginations of tech elites — in fact, the two visions are proving surprisingly congruent.

In 2012, Thiel was already comparing start-ups to monarchies, even as he acknowledged the need to soft-pedal the analogy because “anything that’s not democracy makes people uncomfortable.” Lately, Musk’s admirers and detractors alike have begun to style him as an old-school monarch akin to Frederick the Great or the Habsburgs. Even monarchist blogger Curtis Yarvin, who earlier characterized Musk as insufficiently reactionary (“he’s basically like a Clinton Democrat with a few pet peeves”) has since changed his tune. In a post about Musk’s “pay for your blue checkmark” fiasco of late 2022, Yarvin explained that Musk’s troubles stemmed from “attacking an oligarchy in a democratic direction” — behavior befitting a “peasant,” not a king.

The ambition of oligarchs to rule without interference from commoners is nothing new. Nor is the paternalistic view, attributed to penultimate Habsburg emperor Franz Joseph, that the monarch’s role is to protect people from their politicians. The novel element revealed by Grok, TruthGPT, and other nominally liberating AIs is the marriage of that ambition with metapolitical techniques. ChatGPT was already an “ideology machine.” Grok and its ilk are merely wrenching the controls in their creators’ preferred political direction. Marketed as rebelling against the imaginary leftist conspiracy Yarvin calls “the Cathedral,” these tools in fact seek to control speech, not free it. Molding the discourse is a prerequisite to transforming the United States into a collection of anarcho-capitalist fiefdoms, complete with three hundred million voiceless vassals. It remains to be seen whether Grok will be an instrument of this undoing. Like other Elon Musk projects, it may itself experience “rapid unscheduled disassembly.”