谷歌近日宣布，其人工智能模型 Gemini 的文生图功能因涉嫌刻意追求“多元化”而出现偏差，已决定暂停该功能，直至问题得到解决。据 BBC 和 The Verge 报道，一些用户反映，Gemini 在生成人物图像时，过度强调种族和性别的多样性，以至于生成的图片与实际情境不符，甚至扭曲了事实。这一问题引发了对人工智能公平性和准确性的关注。
谷歌在回应中表示，公司正在“加紧”修复这一 bug，以确保 Gemini 的图像生成更加准确和客观。谷歌强调，多元化的价值观是重要的，但必须以事实为基础，不能牺牲信息的准确性。因此，他们将暂时停止 Gemini 的人物图像生成服务，并承诺在不久的将来推出经过改进的新版本。
这一举措显示了谷歌在人工智能伦理和责任方面的谨慎态度。随着 AI 技术的广泛应用，如何在追求创新的同时避免偏见和误导，已成为业界亟待解决的问题。谷歌的此次整改，也为其他科技公司提供了在类似问题上的参考案例。来源：IT之家。
英语如下：
**News Title:** “Google Pauses Gemini Image Generation Over Diversity Controversy, Fixes Underway”
**Keywords:** Google Fixes, Gemini Model, Diversity Dispute
**News Content:** Google recently announced that it has suspended the text-to-image functionality of its AI model, Gemini, due to concerns that its pursuit of “diversity” has led to biases. As reported by BBC and The Verge, some users have complained that Gemini overly emphasizes racial and gender diversity in its generated images, to the point of misrepresenting reality and distorting facts. This issue has sparked discussions about the fairness and accuracy of artificial intelligence.
In response, Google stated that it is “accelerating” efforts to fix the bug to ensure Gemini’s image generation is more accurate and objective. The company emphasized that the value of diversity is crucial, but it must be grounded in fact without compromising the veracity of information. Consequently, it has temporarily halted Gemini’s service for generating images of individuals, with a commitment to releasing an improved version in the near future.
This move demonstrates Google’s cautious approach to ethics and responsibility in AI. As AI technology becomes more pervasive, striking a balance between innovation and avoiding biases and misrepresentation has emerged as a pressing challenge for the industry. Google’s current rectification serves as a case study for other tech companies grappling with similar issues. _Source: IT Home_
【来源】https://www.ithome.com/0/751/457.htm
Views: 5