OpenAI CEO Sam Altman had a very simple question for xAI CEO Elon Musk's foul-mouthed Grok chatbot: who would make for a "better overall president for the United States?"
The chatbot, which Musk has long positioned as the "anti-woke" alternative to OpenAI's ChatGPT, had an equally simple answer: Kamala Harris.
Grok listed a number of reasons as to why, ranging from "abortion and women's rights" to "inclusivity" and the risk of the disastrous economic policies president-elect Donald Trump has threatened the country with.
In other words, Grok made a compelling case against Trump — the exact kind of argument the extreme right, including Musk, have derided as "woke," a derogatory catch-all for progressive ideals.
"Which one is supposed to be the left-wing propaganda machine again?" Altman asked rhetorically, referring to Musk accusing other AI chatbots like ChatGPT of furthering a progressive agenda.
"We are proud of how consistently ChatGPT scores as the least biased [AI] in evals," he wrote in a follow-up, employing the kind of noncommittal political agnosticism common among tech leaders.
But his jab didn't exactly land — at least not on X-formerly-Twitter. Less than two hours later, Musk snapped back, highlighting a tweet that showed screenshots of Grok refusing to choose between Trump and Harris.
"Choosing between Donald Trump and Kamala Harris as the better president for the United States requires consideration of various aspects including policy, leadership style, and potential impact on the nation and its international relations," Grok wrote, according to the screenshot.
"Swindly Sam is at it again..." Musk wrote, mimicking Trump's penchant for grade school nicknames for his adversaries.
It's unclear if xAI made changes behind the scenes to stop its chatbot from choosing Harris over Trump. But if Altman did get Grok to choose Harris, it's yet another sign that Musk's promises to turn the chatbot into an "anti-woke" conservative propaganda machine haven't worked. We've seen several examples of Grok directly opposing the ideals of the billionaire, from endorsing diversity and inclusion to supporting trans rights.
Earlier this month, Grok even blasted Musk as being a huge spreader of misinformation.
If anything, the spat highlights why it's absurd to let a large language model — a glorified Magic 8 Ball — recommend something as important as a political candidate: the person controlling it can always tweak its answer to flatter their own worldview.
More on Grok: Elon Musk's Grok AI Blasts Elon Musk as Huge Spreader of Misinformation
Share This Article