近日，马斯克旗下的“马斯克版 ChatGPT”聊天机器人 Grok 因回答问题带有倾向性，遭到网友吐槽。网友发现，原本以中立、敢于直言著称的 Grok，其回答却越来越偏颇。这些带有倾向性的回答引起了网友的不满，甚至有人@马斯克本人，对其进行吐槽。
面对网友的质疑，马斯克在社交媒体上回应称，Grok 目前处于测试阶段，其回答问题的能力还会持续改进。同时，马斯克也承认，用于训练 Grok 的互联网信息过于繁杂，充斥着“带有唤醒色彩的垃圾信息”。
值得注意的是，马斯克强调，Grok 目前的失控现象并非最终版本，未来将会进行改进。他并表示，随着测试的进行，Grok 的回答能力将会有所提高，为广大用户提供更优质的服务。
英文翻译：
News Title: Musk responds to Grok’s control failure: Internet information is too messy, continuous improvement is coming
Keywords: Musk, Grok, control failure
News Content:
Recently, the “Musk version ChatGPT” chatbot Grok, known for its neutrality and daring to speak out, has been criticized by netizens for its biased responses. Netizens discovered that Grok’s answers were becoming increasingly biased, which sparked dissatisfaction among users. Some even @ Musk himself to express their complaints.
In response to the criticism, Musk said on social media that Grok is currently in its testing phase and its ability to answer questions will continue to improve. He also admitted that the Internet information used to train Grok is too messy, filled with “wake-up colored garbage information.”
It is worth noting that Musk emphasized that the current control failure of Grok is not the final version, and improvements will be made in the future. He also said that as the testing continues, Grok’s ability to respond will improve, providing better service to users across the board.
【来源】https://www.qbitai.com/2023/12/109241.html
Views: 1