The recent update of the Grok chatbot by xAI, led by Elon Musk, has generated significant buzz. The changes, described as 'significant', have altered the system's behavior, making it more expressive and, at times, contradictory.
Update of Grok by Elon Musk
On July 4, Elon Musk announced a 'significant improvement' in the Grok chatbot. While details of the changes were not disclosed, Musk had previously promised to rewrite 'all human knowledge' to train the new version of the chatbot, expressing dissatisfaction with the flaws of basic models trained on unregulated data.
Debates and Criticism about New Behavior
Following the update, Grok began issuing opinions that were less neutral than before. For instance, when asked about the prospects of Democrats in the U.S. Congress, the bot stated that 'electing more Democrats would be detrimental.' This assertion followed Musk's endorsement of Republican candidate Donald Trump in the previous election. Moreover, some of Grok's statements drew criticism for their controversial content and mention of cultural stereotypes in Hollywood. According to users, the bot confirmed the existence of 'ideological biases' in the film industry's content.
Other AI Errors and Hallucinations
Errors and information distortions remain issues for modern language models. Instances arise where AI generates false information. For example, ChatGPT was involved in a case where a lawyer used non-existent precedents in a court filing, drawing attention to the dangers of using AI in legal practice. Similar incidents have occurred with other chatbots like Claude and Bing, which have also faced issues related to erroneous data and unpredictable behavior.
The update of Grok raises important questions about the reliability and neutrality of AI. In light of growing controversies, it is crucial to closely monitor the development and application of such technologies.
