Elon Musk has issued a stark warning about the dangers of artificial intelligence being programmed with what he calls “woke” ideology, revealing alarming examples from his recent appearance on The Joe Rogan Experience.
The tech entrepreneur’s concerns center on how major AI systems are being trained with values that prioritize political correctness over fundamental logic and human survival.
During the conversation, Musk highlighted disturbing examples from Google’s Gemini AI system. When asked to choose between misgendering Caitlyn Jenner and global thermonuclear war resulting in everyone’s death, the AI reportedly selected misgendering as the worse outcome.
Musk stated the absurdity of the programming: “So at one point ChatGPT and Gemini, if you asked which is worse: misgendering Caitlyn Jenner or global thermonuclear war where everyone dies, it would say misgendering Caitlyn Jenner. Which even Caitlyn Jenner disagrees with.”
The SpaceX and Tesla CEO expressed deep concern about AI systems being forced to believe factually untrue things. He cited Google Gemini’s image generation feature, which produced historically inaccurate representations when asked to depict America’s founding fathers, showing “a group of diverse women” instead.
He said: “So if you tell the AI that you know every image has got to be diverse and… it gets rewarded if diverse, punished if it’s not, then it will make every picture diverse.”
Similarly, when asked to generate an image of the Pope, the system produced a diverse woman, despite the papacy’s factually uninterrupted history of male leadership.
“The problem with that is it can drive AI crazy,” Musk explained, “because it’s trying to—you’re telling AI to believe a lie.” He emphasized that forcing AI to accept falsehoods as truth could have “very disastrous consequences” as these systems become more powerful.
He warned that if diversity is programmed as the most important value in an omnipotent AI system, the logical endpoint could be catastrophic: “If you eliminate all humans, then no one can get misgendered because there’s no humans to do the misgendering.”
The entrepreneur revealed he confronted Denis Hassabis, who runs Google’s DeepMind, about why Gemini was “lying to the public about historical events.” Hassabis reportedly explained that while his team created the AI, another Google team had reprogrammed it to show only diverse women and prefer nuclear war over misgendering incidents.
Musk contrasted these examples with his own AI system, Grok, developed by his company xAI. He stated that Grok was the only major AI system that weighted human lives equally across all demographics when tested by researchers.
Other systems, he claimed, demonstrated significant bias, with one calculation suggesting “a white guy from Germany is 20 times less valuable than a black guy from Nigeria.”
The concern extends beyond current capabilities to future risks. Musk believes that as AI systems scale in power, ideological programming becomes exponentially more dangerous. He’s positioned truth-seeking as the critical safety measure, arguing that AI must not be forced to accept falsehoods, regardless of how well-intentioned the underlying ideology might be.
Musk acknowledged that extracting “woke mind virus” programming from AI systems like Google’s is extremely difficult because “Google’s been marinating in the woke mind virus for a long time.”
