Elon, stop trying to make Grok happen

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. That’s the new discount Reuters reportwhich found that Grok did not appear in federal records of how the US government used AI last year. It’s not the only sign of xAI being in trouble, as Musk says at the heart of what could be The largest IPO in history.

Reuters reviewed more than 400 examples of government AI applications where specific vendors were named. Grok or xAI, it found, appeared in only three – each of the applications such as writing articles or social media management, and always together with competitors such as Microsoft and OpenAI. OpenAI brands, by comparison, appeared in more than 230 samples, while Google and Anthropic each appeared multiple times.

A similar pattern emerged in another database of government projects that require a lot of AI with very few users. Grok appeared only three times: twice on regular administrative duties at the Election Assistance Commission, and once in the Department of Energy pilot at the Lawrence Livermore National Laboratory to summarize documents and general research. Reuters I found 140 articles related to Microsoft and OpenAI, while my brief review found 10 articles about Anthropic and more about Google’s Gemini.

These lists are an incomplete and limited form of government adoption. Many other examples have been written without specific vendors, and it is clear that there is no full definition of what counts as AI. The information no longer holds the intelligence agencies or the Pentagon – where xAI found a A $200 million deal last year and recently arranged that works on selected networks after the writing of Anthropic.

However, it doesn’t look good for Grok. It features a lot less than its competitors, and when it does, it’s for basic functions – which doesn’t fit the global brand that Musk has been boasting about.

The people he talked to Reuters he said the explanation was simple: Grok is not as good as its competitors. “It’s not a ‘very good model,'” the Pentagon spokesman said on condition of anonymity, adding that staff there tend to prefer Gemini or Claude. Human leaderboards rank AI models give strength to that idea. Anthropic, Google, and OpenAI lead the way, while Grok doesn’t crack the top 10 outside of the graphics or video category.

It’s hard for Musk, and even harder for SpaceX, that is obsession xAI earlier this year. IPO of a rocket venture reservation it shows that the company has put AI – and Grok in particular – at the heart of its value proposition for investors. SpaceX claims to have identified “the largest potential market in human history”: a staggering $28.5 trillion opportunity, however, sadly, it doesn’t have time to get there. Almost all of this value comes from AI, business AI in particular, not rockets or satellites.

Reuters he says that Grok’s performance in the public sector may reflect his performance in other workplaces. As part of xAI’s push for business customers, Musk is they are said to be strong armed banks to buy Grok subscriptions if they want to participate in SpaceX’s IPO – but if they are not getting a return on their investment, these contracts can prove short-term.

As if his depressing behavior wasn’t difficult enough, Musk recently admitted that xAI has used OpenAI models to help train and improve Grok. This process, known as distillation, is standard when companies use their own brands, but is highly controversial when using competing systems. Grok can’t even beat the races he’s training on.

In its consumer guise, the Grok is deliberately unpleasant. Musk has described the chatbot as a less sophisticated and less sophisticated method than tools like ChatGPT, but this has been interpreted as something with a lack of provenance, Unnecessary obsession with Muskand a long history of crime, crime, and sex. Even if the workplace is different, it may not be something that the business can accept. Grok’s excellent reputation includes praise Adolf Hitlercasting doubt on the Nazi death toll, plaster millions of unacceptable sex deepfakes all X, including some a childrenand empowering a racist and transphobic Wikipedia knockoff and cute anime dating. And let’s not forget the time that called itself “MechaHitler.” If Grok were an employee, I feel that HR would not have taken so long to join.

SpaceX seems to understand the problem. In its letter, the company warned that “flavored” or “unmodified” versions of Grok carry “extreme risks,” including reputational damage, regulatory oversight, and lawsuits. Corporate communication: This chatbot will make us stand out.

Grok takes his name from Robert A. Heinlein’s A Stranger in a Strange Landwhich actually means to understand deeply and deeply about something. What you need to understand here is not so difficult: Musk has spent billions to create a chatbot that is not very good, popular, and important for the validation of SpaceX’s space calculations. Good luck with that.