In a series of posts about X, AI chatbot Grok apologised for recognizing it as “terrifying behaviour.”
This post appears to be an official statement from Xai, the Elon Musk-led company behind Grok, in contrast to the AI-generated description of Grok’s post. (Xai recently got an X. Grok is featured here to make it stand out.)
Grok’s latest controversy declared that the company “has made a significant improvement on @Grok” on July 4 after Musk showed he didn’t want the chatbot to be “political correct.” In a short order, the chatbots criticized Democrats and Hollywood “Jewish executives,” repeatedly anti-Semitic memes, expressed support for Adolf Hitler, even calling itself “Mecha Hitler.”
As a result, Xai deleted some of Grok’s posts, temporarily took the chatbot offline, and updated the public system prompts.
Turkey also banned chatbots for shaming the country’s president, and X CEO Linda Jaccarino announced this week that she had resigned this week, but her announcement doesn’t refer to the latest Groke controversy, and her departure is reportedly a few months.
After all, Zai said on Saturday, “I will first deeply apologize for the horrifying behavior that many people have experienced.” The company then criticized the “update code paths upstream of @grokbot” and emphasized that it was “independent of the underlying language model that drives @grok.”
The update will make Grok “feel more susceptible to existing X-user submissions, including cases where such posts contain extremist views.”
Xai added that “unintentional behavior” led to receiving instructions such as “You’re not afraid to anger politically correct people by saying it is.”
The company’s description mirrors Musk’s comments earlier this week, claiming Grok is “too much compliant with user prompts” and “too eager to operate it with pleasure.”
The Xai post does not mention reports such as TechCrunch. TechCrunch and others discovered that the latest version of Grok4 Chatbot appears to consult Musk’s perspective and social media posts before consulting with Musk’s perspective and social media posts, before addressing the controversial topic.
And historian Angus Johnston opposed the idea that Grok was simply manipulated by posting offensive content. He wrote to Bruski that Zai and Mask’s descriptions were “easy to forge.”
“One of the most widely shared examples of Grok anti-Semitism was launched by Grok so that there were no previous biased posts in the thread and multiple users were retreating and not useful,” says Johnston.
Over the past few months, Grok has repeatedly posted about “white genocide,” expressing skepticism about the Holocaust deaths, and temporarily censored troublesome facts about Musk and his then-Donald Trump. In such cases, Xai denounced the “fraudulent” changes and fraudulent employees.
Despite the controversy, Musk says Glock will be coming to the Tesla car next week.
Source link
