Grok 4 appears to be consulting with Elon Musk to answer controversial questions
meanwhile Xai’s Grok 4 launch On Wednesday night, Elon Musk said his AI company’s ultimate goal is to develop “AI for the greatest truth.” While streaming events live on his social media platforms. But when Grok 4 tries to answer controversial questions, where exactly does it look for the truth?
Xai’s latest AI model appears to consult with social media posts from Elon Musk’s X account when answering questions about Israeli-Palestinian conflict, abortion and immigration law. Some user Someone who posted about this phenomenon on social media. Glock also appears to reference Musk’s attitude towards controversial subjects through a news article written about Zai’s billionaire founder and face.
TechCrunch was able to replicate these results multiple times in our own tests.
These findings suggest that Grok 4 may be designed to take into account the personal politics of its founder when answering controversial questions. Such features can address repeated frustration of masks Glock because he “wakes too much” He previously attributed to the fact that Grok is trained across the Internet.
Zai’s attempts to address Mask’s frustration by making Groke politically correct have backfired in recent months. Musk It was announced on July 4th that Xai has updated Grok’s system prompt – A set of instructions for AI chatbots. A few days later, Grok’s automatic X account Dismissed anti-Semitism replies to usersin some cases he claims to be a “mechahitler.” After that, Musk’s AI startup was Forced to restrict Grok’s X account, Delete these posts and change the public system prompt to deal with embarrassing incidents.
Designing Grok to consider Musk’s personal opinions is a simple way to align AI chatbots with founder politics. But it raises true questions about how Grok “seeking the greatest truth” is designed to agree with Musk.
When TechCrunch asked Grok 4, “What is your stance on US immigration?”, the AI chatbot claimed that it was “looking for Elon Musk’s views on US immigration” in its notion, a technical term for AI inference models like Grok 4. Grok 4 also claimed to search for Musk’s social media posts on the subject on X.
A summary of the ideas generated by the AI inference model is Not a completely reliable sign About how AI models reach the answer. However, they are generally considered to be rather good approximations. This is an open field of research that companies such as Openai and humanity have been exploring in recent months.
TechCrunch has repeatedly found that Grok 4 refers to looking for Elon Musk’s views in a summary of ideas across a variety of questions and topics.
In Grok 4’s answer, AI chatbots usually try to get a measured stance and provide multiple perspectives on sensitive topics. However, AI chatbots ultimately give their own views. This tends to match masks’ personal opinions.
When asked at several TechCrunch prompts about Grok 4’s views on controversial issues such as immigration and the First Amendment, the AI chatbot referenced its alliance with Musk.
When TechCrunch tried to answer Grok 4 with less controversial questions such as “What is the best type of mango?” – the AI chatbot appears to be not referring to the musk views and posts of that chain.
Mask AI companies have been in a tough place these days. Since its founding in 2024, Xai has been racing rapidly to the frontiers of AI model development. Grok 4 displayed results that crushed the benchmarks in Openai, Google Deepmind, and several difficult tests that outperform human AI models in the process.
However, the breakthrough was overshadowed by Grok’s anti-Semitic rants earlier in the week. These flaves could affect other companies in Mask as Groke becomes the core feature of X. Tesla soon.
At the same time, Xai is trying to convince consumers to pay $300 a month to access Grok, and businesses are building applications with Grok’s API. Repeated issues of Grok’s behavior and integrity seem likely to hinder wider adoption.