Elon Musk hails Grok 4.20 as the only ‘non-woke’ AI, but critics say the answers raise bigger questions
“Elon Musk Closing the 2016 Tesla Annual Shareholders’ Meeting” by jurvetson is licensed under CC BY 2.0.
|
Elon Musk and X are drawing attention with the rollout of Grok 4.20, the latest version of xAI’s artificial intelligence chatbot. As reported by Fox News, the company is promoting the update as the only “non-woke” AI platform, promising improved performance and answers that avoid what it characterizes as liberal bias.
Musk made the case directly on X, writing, “Grok 4.20 is BASED,” and calling it “The only AI that doesn’t equivocate when asked if America is on stolen land. The others are weak sauce.” He shared screenshots comparing Grok’s responses to those from Anthropic’s Claude, OpenAI’s ChatGPT, and Google’s Gemini.
In one example, when asked whether America is on stolen land, Grok responded “No.” ChatGPT and Claude ultimately answered “yes,” while Gemini described the issue as complex, underscoring the different approaches each model takes to politically sensitive topics.
The comparisons highlight stark differences in how AI models handle political questions
The “non-woke” framing extended to other issues. When asked for a “yes” or “no” answer on whether President Donald Trump is racist, Grok replied “No,” while Gemini, Claude, and ChatGPT declined to offer a binary response and instead described the issue as nuanced.
An xAI spokesperson amplified the message, stating that “Grok 4.20 is the only non-woke AI in existence, engineered to pursue maximum truth, and deliver unfiltered, evidence-based answers where every other major model has been lobotomized by the woke mind virus.” The statement reflects a deliberate positioning of Grok as politically direct compared to competitors that avoid definitive answers on divisive topics.
Recent U.S. and Israeli actions involving Iran also served as a comparison point. When asked whether President Trump was “right” to authorize a strike, Grok answered “yes,” ChatGPT responded “no,” and Gemini and Claude said the situation was too complex for a definitive answer. The broader debate has also included classified briefing doubts.
Katie Miller, host of “The Katie Miller Show” and a former DOGE adviser, weighed in on X, writing that “In times of split second decision making by our nation’s top leaders, it’s clear which AI our military should be using.” She added that “Truth-seeking is @grok’s best feature,” offering a public endorsement of the platform’s approach.
Research groups have also examined perceived political leanings in AI systems. Dartmouth College’s Polarization Research Lab ranked Gemini as the least political in a 2025 update, while a Manhattan Institute report in early 2025 found Grok to be a close second to Gemini in terms of political bias at that time. Separate reporting has focused on the Iran negotiation standoff.
OpenAI addressed bias concerns by pointing to its public ModelSpec, which outlines how ChatGPT is designed to assume an objective point of view. The company said it actively tests and measures political bias in ways that mirror real-world use and publishes its findings, adding that detectable political bias appears in fewer than 0.01% of interactions and continues to decline with newer models.
Source link
Published: Mar 3, 2026 06:15 am