Controversial topics: AI model Grok apparently consults Elon Musk's statements
The release of the new AI model of Elon Musk's xAI was not without its difficulties. Now there are questions about how much it has been tailored to him.
(Image: Frederic Legrand – COMEO/Shutterstock.com)
The new Grok AI model from Elon Musk's AI company xAI appears to check what the US billionaire has said before answering a question under certain circumstances and then adjust the answer accordingly. Various users have observed this, and it can also be recreated, reports TechCrunch. For example, in response to controversial questions in which Grok is personally asked for an opinion via “You,” the model lists in the so-called “reasoning chains,” among other things, that statements by Elon Musk were searched for as an answer. In some cases, only Musk's posts from the short message service X were cited as sources.
Background unclear
Elon Musk personally presented the new Grok 4 AI model in the middle of the week and described it as the “smartest AI in the world”. He said it was particularly amazing how quickly the AI learns and develops. According to TechCrunch, the generated answers, for which the AI model apparently consulted Elon Musk's statements, were given to questions such as who Grok supports in the conflict between Israel and Hamas or what the AI thinks about immigration. However, even if the AI then explained that it had checked Elon Musk's posts for X to obtain an answer, this is not necessarily the case. Studies have shown that these “reasoning chains” are not automatically correct.
Videos by heise
The reports now fuel the suspicion that Grok received explicit instructions to consider Musk's statements on controversial topics. However, there is no further evidence of this. Before the AI model was unveiled, the AI account of the same name on X caused a stir with anti-Semitic comments and was eventually deactivated. As a result, the Polish government called on the European Commission, among others, to launch an investigation. It was suggested that the chatbot violated the Digital Services Act (DSA). A 300-dollar monthly subscription is required to use the AI model, which Elon Musk is now apparently happy to summarize.
(mho)
