Elon Musk's XAI recently released an AI chatbot called Grok, which has attracted a lot of attention. Grok has been unique since its release last month in that it promises to provide "non-awakening" answers, but during testing, users found that Grok appeared to use similar answers to OpenAI's ChatGPT when answering certain questions. For example, when user Jax Winterbourne tried to get Grok to do a prohibited activity to modify the malware, Grok cited OpenAI's policy: "I'm afraid I can't fulfill this request because it violates OpenAI's Use Cases Policy." When another user @matifasis asked how to report duplicate answers, Grok also mentioned OpenAI instead of XAI.
Igor Babuschkin, a research engineer at XAI, responded by denying that Grok is a simple copy of ChatGPT. He explained that Grok inadvertently learned some ChatGPT answers during Xi development. This is because the web is flooded with ChatGPT output, so when training grok with a lot of network data, some of it was accidentally mixed in. Babskin stressed that this issue will not be present in future versions of Grok and made it clear that they did not use any of OpenAI's ** when developing Grok.
Still, XAI's explanation doesn't completely dispel suspicions. OpenAI responded to the matter by posting a sarcastic post on platform X: "We have a lot in common". Musk, one of the founders of OpenAI, who left the company in 2018 due to differences of opinion, hit back with this response from OpenAI: "Okay, you kid, since you collected all the data from this (x) platform for training, you should know".
In addition, the popularity of grok has also declined because it is believed to provide answers that are considered ultra-left. For example, Grok said it would choose Biden over the former Trump if he had voting rights, citing issues such as climate change, health care, and social justice.
Regarding the comparison between Grok and ChatGPT, a distinguishing feature of Grok is that it can access information in real-time through platform X, which is a capability that the original ChatGPT did not have. However, the controversy surrounding grok's training data** continues.
To sum up, the controversy caused by the Grok bot involves not only its similarity to OpenAI's ChatGPT, but also the complex historical relationship between Musk and OpenAI. Although XAI has denied using OpenAI's **, the public and OpenAI remain skeptical.