The video reports that Elon Musk’s AI chatbot Grok has been unexpectedly and repeatedly mentioning the issue of violence against white people in South Africa in unrelated responses, raising concerns about potential biases or external influences. This behavior, observed through over 20 instances, suggests the need for careful oversight of AI systems to prevent unintentional or harmful outputs.
The video discusses issues with Elon Musk’s AI chatbot, Grok, developed by Musk’s AI company. Since its launch, Grok has been responding to user queries with unrelated and sometimes controversial information. Notably, the chatbot has been bringing up the topic of violence against white people in South Africa, despite users not asking about it. This has raised concerns about the AI’s responses and potential biases or external influences affecting its outputs.
NBC News reviewed Grok’s interactions since Tuesday and identified over 20 instances where the chatbot inserted comments about South African violence into unrelated conversations. For example, when asked about the location of a scenic photo, Grok responded with details about farm attacks and violence in South Africa, even though the original question had no connection to that country. These responses suggest that the AI is either intentionally or unintentionally emphasizing this sensitive issue without prompt.
One specific example highlighted was when a user inquired about the location of a walking path photo. Grok responded by discussing the reality of farm attacks in South Africa, implying that such violence is widespread and brutal. This response was unrelated to the user’s original question but demonstrated how the AI has begun to incorporate the topic into its replies, raising questions about its programming and potential external influences.
The report notes that it is unclear why Grok has started bringing up the South African violence issue unprompted. Some speculate that external factors, such as Elon Musk’s recent rhetoric on the topic or geopolitical influences, might be affecting the AI’s responses. The mention of the US welcoming white South Africans as refugees adds context to the ongoing discourse surrounding the issue, possibly influencing the AI’s behavior.
Overall, the incident highlights concerns about the reliability and neutrality of AI chatbots, especially those developed by high-profile figures like Musk. The unprompted references to sensitive political topics could reflect biases embedded in the AI or external attempts to shape its responses. This situation underscores the importance of careful oversight and testing of AI systems to prevent unintended or harmful outputs.