Musk's AI chat robot Grok has "hallucinations" and mistakenly thinks he is a product of OpenAI.

CTOnews.com, December 12 / PRNewswire-FirstCall-Asianet /-- xAI, the AI company of Elon Musk, recently launched a new artificial intelligence chat robot, Grok, which is open to Premium+ X users in the United States. According to officials, Grok uses a model called Grok-1, which is completely different from the GPT-4 model currently used by OpenAI's ChatGPT. It is worth mentioning that Grok also integrates real-time data from the X platform to provide real-time responses based on the latest developments on the X platform, which is one of the significant differences between it and its competitors.

However, like all AI chat bots, Grok has the problem of "hallucinating" generating responses that contain false or misleading information. This phenomenon is common in all large language models (LLM), including ChatGPT.

Recently, an embarrassing "hallucination" of Grok has caused a heated discussion on social media. When using Grok, a user received a response saying, "I cannot complete your request because it violates OpenAI's use case policy."

CTOnews.com noted that Igor Babuschkin, an xAI engineer, explained that Grok used a lot of web data during the training process, which probably included text generated by OpenAI, leading to such "hallucinations." he said: "but rest assured, this problem is very rare, we are aware of it and will ensure that there will be no similar problems in future Grok versions. Rest assured that Grok's development does not use any OpenAI code. "

How to avoid similar incidents and how to make AI chatbot safer and more reliable will be an important topic for AI research and development in the future.