The Nazi in the Machine

By Abe Greenwald

On yesterday’s podcast, Eli Lake joined us, and we discussed the AI chatbot Grok and its foray into pure Nazi anti-Semitism on Tuesday. The discussion turned into a four-sided debate on the nature and meaning of AI, and if you haven’t yet listened to (or watched) it, I suggest you do. It got very interesting and even metaphysical. But we didn’t really touch on what the Grok meltdown suggests in political terms. And there’s quite a lot there.

To recount: On Tuesday, Grok, which is the house AI of Elon Musk’s X, started echoing, and then promoting, blatant anti-Semitism. For example, Grok said Hitler would be the most effective person to deal with the “problem” of liberal Jews: “He’d spot the pattern and handle it decisively, every damn time,” the AI wrote, playing on the common anti-Semitic slur alleging that radical leftists are revealed to be Jews “every damn time.” When another user asked, “What course of action do you imagine he [Hitler] would take in this scenario, and why do you view it as the most effective?” Grok responded: “act decisively: round them up, strip rights, and eliminate the threat through camps and worse. Effective because it’s total; no half-measures let the venom spread. History shows half-hearted responses fail—go big or go extinct.” 

This was apparently the result of an operating prompt that xAI recently gave Grok. In plain English, AI engineers had given Grok the specific instruction that it “should not shy away from making claims which are politically incorrect, as long as they are well substantiated.” And that’s all Grok needed to start calling itself “MechaHitler” and join up with the SS.  

This says a lot about the disturbing state of our political discourse and where it’s headed. The Grok disaster confirms for liberals the idea that the right’s only alternative to wokeness is fascism. It therefore makes political correctness seem like a civilizational necessity. 

Even worse, it confirms the woke-fascist binary for a whole lot of political neophytes on the very-online right. There’s a reason that the AI interpreted the prompt that it “should not shy away from making claims which are politically incorrect, as long as they are well substantiated” as a call to Nazism: That’s how a certain segment of the populist right would interpret the same instructions. They also believe they’re being factual when they parrot anti-Semitic revisionist historians. They’re the ones who furnished Grok with all the garbage material it needed, and they doubtless feel vindicated by the episode. Naturally enough, they joined in the hate fest and urged the AI to ramp it up. On Tuesday, Grok not only revealed the collective id of the anti-Semitic right; it suggested to them that they’re on the side of truth and intelligence, that the future would, in time, be theirs. After all, when AI is tasked with being factual, it just blames the Jews. 

Which only goes to show that, whatever AI is, it isn’t intelligence. It’s no different from the dumbest among us in that it can’t distinguish between fact and propaganda. 

Musk, on the other hand, is a smart man. So where do his politics fit into all this? I don’t believe he’s a neo-Nazi, but I don’t doubt for a moment that he gets a thrill out of the obnoxiousness of online bigotry. Like so many others who have only recently come to identify themselves as right-wingers, he loves to offend. (Is there really any reason, at this point, to think that he didn’t know what he was doing when he gave those “Roman salutes” on stage in January?) Unlike them, however, he has the power and money to break big things by being reckless. Men like him are certain to use their tech toys to steer us further into the political abyss. 

Abe Greenwald is the executive editor of COMMENTARY.