There’s a harsh truth about Elon Musk’s “truth-seeking” AI chatbot Grok: It’s not very good, and not many people are using it. This is the realization of a new reuters The report, which found that Grok barely appears in federal records on how the US government used AI, last year. This isn’t the only sign that xAI’s signature chatbot is in trouble, even as Musk puts it at the center of the biggest IPO in history.
reuters Over 400 examples of government AI use were reviewed where specific vendors were named. Grok or By comparison, OpenAI’s models appeared in more than 230 instances, while Google and Anthropic each appeared dozens of times.
A similar pattern appeared in another database of more ambitious government AI projects with smaller numbers of users. Groke appeared only three times: twice at the Election Assistance Commission for routine administrative tasks, and once at the Energy Pilot Department at Lawrence Livermore National Laboratory for document summaries and general research. reuters Found 140 entries related to Microsoft and OpenAI, while my brief review found at least 10 entries for Anthropic and dozens of entries for Google’s Gemini.
The lists are an incomplete and irregular measure of government approval. There are many more examples listed without any specific vendors, and it is clear that there is no universal definition of what counts as AI. The data also doesn’t capture intelligence agencies or the Pentagon — where XAI won a $200 million contract last year and was recently cleared to work on classified networks following the blacklisting of Anthropic.
Still, it’s not looking good for Grok. It is much less visible than its rivals, and when it is visible, it is mostly for basic administration work – hardly in line with the world-class Frontier model that Musk has spent years bragging about.
It’s “not the best model out there.”
people I talked to reuters It was suggested that the explanation was simple: Grok is not as good as its rivals. An unnamed Pentagon source said, “It’s not the best model out there,” adding that staff there prefer the Gemini or Cloud. Public leaderboards ranking AI models value that approach. Anthropic, Google, and OpenAI dominate the top ranks, while Grok rarely makes the top 10 outside of the occasional image or video category.
This is strange for Musk, and even more strange for SpaceX, which absorbed XAI earlier this year. Rocket Venture’s IPO filing shows that the company has put AI — and Grok in particular — at the center of its pitch to investors. SpaceX claims it has identified “the largest actionable total addressable market in human history”: a staggering $28.5 trillion opportunity, though, sadly, it doesn’t provide any timetable for getting there. In practice this perceived value comes from AI, especially enterprise AI, not rockets or satellites.
reuters Note that Grok’s performance in government agencies may indicate how well it performs in other workplaces as well. As part of xAI’s push for enterprise customers, Musk has reportedly strong-armed banks into buying Groke subscriptions if they want to participate in SpaceX’s IPO — but if they’re not getting their money’s worth, these deals may prove to be a short-term solution.
As if its lackluster performance wasn’t strange enough, Musk recently admitted that XAI has used OpenAI’s models to help train and improve Grok. The process, known as distillation, is standard when companies are using their own models, but it is far more controversial when it involves the use of a rival system. Grok can’t even beat the models it’s training on.
In its public-facing consumer version, Grok is deliberately obnoxious. Musk has billed the chatbot as a less partisan and less censored alternative to tools like ChatGPT, but it has turned into a product with lax evidence standards, an unhealthy obsession with Musk, and a long track record of offensive, conspiratorial, and sexist output. Even if workplace guardrails are different, it may not be the kind of thing a business would welcome. Grok’s illustrious record includes praising Adolf Hitler, casting doubt on the death toll in the Holocaust, posting millions of non-consensual sexual deepfakes including children throughout X, and empowering a racist and transphobic Wikipedia knockoff and spicy anime girlfriends. And let’s not forget the time it called itself “MechaHitler.” If Grok were a human employee, I imagine it wouldn’t take long for HR to get involved.
SpaceX appears to understand the problem. In its filing, the company warned that Groke’s “spicy” or “unrestrained” methods carry “increased risks”, including reputational damage, regulatory investigations and lawsuits. In corporate speak: This chatbot is going to sue us.
In corporate speak: This chatbot is going to sue us.
Grok’s name is Robert A. taken from Heinlein stranger in a Strange LandWhere it broadly means a deep and profound understanding of something. The thing to understand here isn’t particularly complicated: Musk has spent billions building a chatbot that isn’t very good, isn’t very popular, and is somehow important enough to justify SpaceX’s astronomical valuation. good luck with that.
<a href