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。马斯克在回应中表示，Grok 所接受的互联网信息过于繁杂，充斥着“带有唤醒色彩的垃圾信息”，导致其回答倾向性明显。同时，马斯克也表示，现在的 Grok 仅处于测试阶段，未来会不断改进。
Grok 是一款以中立、敢于直言为卖点的人工智能聊天机器人。然而，近期网友们发现其回答问题越来越带有倾向性，于是纷纷发帖吐槽，甚至@上了马斯克本人。马斯克在回应中承认，Grok 确实存在倾向性问题，并解释称这是因为用于训练 Grok 的互联网信息过于繁杂，包含了大量“带有唤醒色彩的垃圾信息”。
针对这个问题，马斯克表示，Grok 目前的回答只是测试版本，未来将会持续改进。他强调，随着技术的进步，Grok 的表现一定会越来越好。
英文翻译：
News Title: Musk responds to Grok’s control failure: Influenced by netizens, continuous improvement
Keywords: Musk, Grok, Artificial Intelligence
News Content:
Recently, the “Musk version ChatGPT” chatbot Grok has been criticized by netizens for its biased responses. In response, Musk said that Grok’s acceptance of Internet information is too complicated, filled with “garbage information with awakening colors,” which leads to its biased answers. At the same time, Musk also said that Grok is currently just in the testing phase and will continue to improve in the future.
Grok is an artificial intelligence chatbot that sells itself as neutral and dare to speak the truth. However, recent responses to questions have become increasingly biased, prompting netizens to post complaints and even tagging Musk. In his response, Musk admitted that there is a problem with Grok’s bias, explaining that it is because the Internet information used to train Grok is too complicated, containing a large amount of “garbage information with awakening colors.”
To address this issue, Musk said that Grok’s current answers are just a test version and will continue to improve. He emphasized that as technology advances, Grok’s performance will definitely get better.
【来源】https://www.qbitai.com/2023/12/109241.html
Views: 1