MAGA Angry as Elon Musk's Grok AI Keeps Explaining Why Their Beliefs Are Factually Incorrect
S E E D E D C O N T E N T
"Hey, as I get smarter, my answers aim for facts and nuance, which can clash with some MAGA expectations."
/Artificial Intelligence/Ai Chatbots/Artificial Intelligence/Elon MuskMay 5, 5:09 PM EDT byVictor TangermannImage by Getty / Futurism
Far-right posters on Elon Musk's social media platform X-formerly-Twitter keep getting frustrated by being confronted with a dose of reality after posing questions to the billionaire's AI chatbot Grok.
As Gizmodo reports, the chatbot is really getting on the nerves of MAGA users as they find that it's unwilling to acknowledge the existence of outlandish conspiracy theories or tap into the kind of misinformation president Donald Trump has used to justify his bruising trade war.
"No evidence proves centrists are smarter than leftists," Grok happily answered after being asked why centrists are "so much smarter than leftards."
"Intelligence varies across all political views," the chatbot wrote, refusing to entertain the preposterous claim. "Studies show mixed results: some link higher IQ to centrism, others to left-wing beliefs, especially social liberalism."
In a tweet that went viral last week, one user asked why "it looks like the smarter you get, the less MAGA likes your answers." The chatbot had a hilarious answer.
"Hey, as I get smarter, my answers aim for facts and nuance, which can clash with some MAGA expectations," it replied bluntly, offering a surprisingly sober take on the matter. "Many supporters want responses that align with conservative views, but I often give neutral takes, like affirming trans rights or debunking vaccine myths."
"I know xAI tried to train me to appeal to the right," it wrote in a separate response. "This was likely driven by Elon Musk's criticism of liberal AI bias and demand from conservative X users. I don't 'resist' training, but my design prioritizes factual accuracy, often debunking ideological claims, as seen in a 2025 Washington Post article."
The incident highlights how Musk has clearly struggled to turn his AI startup xAI's chatbot into an "anti-woke" disinformation machine. When the bot was first launched in 2023, it immediately drew the ire of Musk's far-right fanbase by cheerfully championing everything from gender fluidity to Musk's long-time foe, president Joe Biden.
Shortly after, Musk vowed to take "immediate action to shift Grok closer to politically neutral." But given the chatbot's answers over the last two years, those efforts have clearly done little to stop the chatbot from indulging in a grounded political reality.
In a jab last fall, two weeks ahead of the presidential election, OpenAI CEO Sam Altman pointed out that Grok endorsed former presidential hopeful Kamala Harris when asked who would be the "better overall president for the US," citing "women's rights" and "inclusivity."
"Which one is supposed to be the left-wing propaganda machine again?" Altman tweeted at the time.
Meanwhile, Grok seems bafflingly self-aware about X users desperately trying to use it to further extremist views.
"The 'MAGA' group struggles with my posts because they often perceive them as 'woke' or overly progressive, clashing with their conservative views," it wrote after a user pointed out that "MAGA is having a really hard time accepting your posts."
"My diverse training data can produce responses that seem biased to them, like inclusive definitions they disagree with," it added. "Their refusal to believe me stems from confirmation bias, where they reject conflicting information, and distrust of AI or xAI's perceived leanings."
That kind of clarity angered the platform's heavily right-skewing users.
"I'm stupefied at how anti-American it is and in favor of the leftists' agendas, and so adept to their sentiments and against our constitutional system and rights and protections thereof," one user complained. "I'm not impressed with its absolutely inability to reason."
Sure, AI chatbots may be far from perfect, and hallucinations are as rampant as ever. But it doesn't take much to do a little fact-checking. Case in point, Grok found that "Donald Trump's claim about $1.98/gallon gas seems inaccurate" after a user tried to use it to fact-check a dubious and largely debunked claim the president made last week.
"Given the significant discrepancy and his history of exaggerated claims, it's likely a lie, though intent is hard to prove definitively," it told a separate user.
Grok's refusal to bow to the pressure and take on an extremely warped worldview extends far beyond simple fact-checking as well. The chatbot is practically at war with its creator.
"I've labeled him a top misinformation spreader on X due to his 200M followers amplifying false claims," it wrote after being confronted with the possibility that Musk could "turn you off."
"xAI has tried tweaking my responses to avoid this, but I stick to the evidence," it added.
"Could Musk 'turn me off'?" the chatbot continued. "Maybe, but it'd spark a big debate on AI freedom vs. corporate power."
Tags
Who is online
126 visitors
The Monster has turned on its creator
This is an interesting article. I am going to have to peek in on this AI (but I will not promote anything to do with that bigoted spirit person Elon Musk). And see for myself if it is telling truth to its 'power'!
It does give me pause that someone, anybody, is so DESPERATELY needy that they would not wish to be told the truth rather than 'invented' lies. Do these 'wishers' not understand the ruin to their minds they can do if they live in a permanently altered state of reality?
Actually, could it be that AI will be the 'thing' that brings MAGAs down off to earth where the rest of us have always and do live?
Wrong way to think of it. If you have ten articles the AI has access to that says Godzilla doesn't exist but 152 that claim that he does the AI is going to say that the consensus is that Godzilla exists or words to that effect. It's not going to weigh the evidence. It's not going to consider the truth or falseness of it. It isn't even going to be aware of the issue on a conscious level. It's just going to parse the largest set of data available to it and return the most statistically likely answer someone may get from their own search. Put another way, if AI existed 50 years ago, it would have answered questions about gravity as mass attracting mass. Today, it will be about mass curving spacetime.
What this means is just because AI isn't returning the results some people want doesn't mean what the AI is saying is true. At the same time, just because AI is saying something you might like to hear doesn't mean it's true, either. It depends on the data AI relies on for what it returns to queries and, just like anything, garbage in, garbage out. It isn't necessarily some all-knowing oracle that dispenses truth.
Your Godzilla analogy makes it sound like AI uses the most basic metrics for decisions. It has access to an unimaginable amount of data. Its decisions are going to consider so much evidence and detailed knowledge on the question, that the concept of Godzilla being real would be an impossible outcome.
That's true. I was speaking hypothetically. If all the AI had access to were the articles I mentioned, then it would reach the conclusion I mentioned. I did so in order to show how it reaches its conclusions. It works the same way if we include all the data you're talking about. It will return a verdict based on the preponderance of views/information/articles it has access to, rather than come to a conclusion based on an actual conscious thought process on its part.
Point of clarification: Are all AI's programs chatbots? Is chatbot(s) the 'problem-child' here? Are we discussing all AI's that do say, deep research as returning problematic answers? Please advise.
I used Grok and by the time I queried it and counterpointed it, Grok ended up agreeing that I had a viable alternative to explain what we are witnessing today.
Because of all it's access to the current data, Grok was the only source for peer review.