据BBC和The Verge近日报道，谷歌正紧急处理其人工智能模型Gemini的一项技术问题。有用户反映，Gemini在生成文本相关的图像时，存在过度追求“多元化”的现象，以至于在某些情况下，生成的图像并未准确反映实际情境，出现了性别和种族的不实表现。这一情况引发了关于模型公正性和准确性的讨论。
为了回应这些关切，谷歌已决定暂时停止Gemini的人物图像生成功能，以便对这一“多元化”倾向的bug进行修复。谷歌在一份声明中表示，他们认识到这一问题的重要性，并承诺将尽快推出改进后的版本，以确保图像生成更加准确、公正且无偏见。
此次事件凸显了科技公司在推动多元化的同时，也需谨慎处理算法可能产生的偏见问题。谷歌的这一举措，旨在维护其产品在用户中的信任度，并继续致力于提供准确、包容的AI服务。我们将持续关注Gemini的更新情况，以及谷歌在解决此类问题上的进展。
英语如下：
News Title: “Google Pauses Gemini Text-to-Image: Pursuing Diversity at the Cost of Accuracy, Restarting After Fix”
Keywords: Google fix, Gemini controversy, text-to-image suspension
News Content:
Title: Google Temporarily Halts Gemini Text-to-Image Feature Over Excessive ‘Diversity’ Concerns
Recent reports from BBC and The Verge reveal that Google is urgently addressing a technical issue with its AI model, Gemini. Users have noted that the model, while generating text-related images, overly emphasizes “diversity,” sometimes resulting in images that inaccurately depict the context, displaying incorrect gender and racial representations. This has sparked discussions about the model’s fairness and accuracy.
In response to these concerns, Google has decided to suspend Gemini’s functionality for generating character images to fix this “diversity” bias. The company stated in a press release that it acknowledges the issue’s significance and pledged to roll out an updated version promptly, aiming to ensure more accurate, fair, and unbiased image generation.
This incident highlights the challenge tech companies face in promoting diversity while also mitigating potential biases in their algorithms. Google’s move is intended to maintain user trust and its commitment to providing accurate and inclusive AI services. We will continue to monitor Gemini’s updates and Google’s progress in addressing such issues.
【来源】https://www.ithome.com/0/751/457.htm
Views: 1