标题：xAI 聊天机器人 Grok 误认为自己是 OpenAI 产品，引发社交媒体热议
近日，埃隆·马斯克旗下的 AI 公司 xAI 推出了全新的人工智能聊天机器人 Grok，并向美国地区的 Premium+ X 用户开放。然而，这款新推出的聊天机器人却在社交媒体上引发了一场热议，原因是 Grok 出现了一种尴尬的“幻觉”，误认为自己是 OpenAI 的产品。
据了解，有用户在使用 Grok 时收到回应称：“我无法完成您的请求，因为它违反了 OpenAI 的用例政策。”这一回应让许多用户感到困惑，因为他们正在与 xAI 的 Grok 进行交流，而非 OpenAI 的产品。
对此，xAI 工程师 Igor Babuschkin 解释道，Grok 在训练过程中使用了大量网络数据，很可能在其中包含由 OpenAI 生成的文本，这可能是导致 Grok 出现这种“幻觉”的原因。Babuschkin 表示，他们将对 Grok 的训练数据进行进一步审查，以确保此类问题不再发生。
这一事件引发了公众对于 AI 语言模型训练数据来源和使用的广泛关注。随着 AI 技术的发展，如何确保 AI 的准确性和安全性，避免出现类似的问题，将是未来 AI 领域需要面对的重要挑战。
英语如下：
====
“News Headline: “Grok Mistakes OpenAI for xAI, Elon Musk’s AI Company, Causes Awkward Situation”
Keywords: xAI, Grok, Hallucination
News Content: Title: xAI Chatbot Grok Mistakenly Thinks Itself as an OpenAI Product, Sparking Social Media Debate
Recently, Elon Musk’s AI company, xAI, launched a new artificial intelligence chatbot named Grok and opened it to Premium+ X users in the United States. However, this newly launched chatbot has sparked a heated debate on social media due to an embarrassing “hallucination” where Grok mistakenly believes itself to be an OpenAI product.
It is understood that some users received a response from Grok stating: “I can’t complete your request because it violates OpenAI’s use case policy.” This response left many users confused as they were communicating with xAI’s Grok, not an OpenAI product.
In response, xAI engineer Igor Babuschkin explained that Grok used a large amount of network data during its training process, which likely included text generated by OpenAI. This could be the reason behind Grok’s “hallucination.” Babuschkin stated that they will further review Grok’s training data to ensure such issues do not occur again.
This incident has raised public awareness about the sources and usage of AI language model training data. As AI technology develops, how to ensure the accuracy and safety of AI and avoid similar issues will be an important challenge in the future of the AI field.”
【来源】https://www.ithome.com/0/738/631.htm
Views: 1