谷歌近日宣布，由于其 Gemini 大模型的文生图功能被曝出存在刻意追求“多元化”的技术问题，公司将暂停该功能，以进行紧急修复。这一消息源自 BBC 和 The Verge 的综合报道，指出 Gemini 在生成人物图像时，可能会过度强调性别和种族的多样性，以至于生成的结果与实际情况不符，甚至忽略史实。
据用户反馈，Gemini 在生成图片时，倾向于展示各种性别和种族的人物，这一现象引发了关于图像真实性和算法偏见的讨论。谷歌对此表示重视，决定立即采取行动，以消除可能存在的偏见，并确保其人工智能技术的公正性和准确性。
谷歌发言人表示，公司将尽快修复这一问题，并重新发布改进后的 Gemini 图像生成功能。这一举措彰显了谷歌对于消除技术偏见、维护用户信任的承诺。随着人工智能在日常生活中的广泛应用，确保算法公正无偏已经成为业界亟待解决的重要议题。谷歌的此次行动，无疑为行业树立了一个积极的榜样。
英语如下：
Title: “Google Pauses Gemini Image Generation over Diversity Bias: Urgent Fix Underway”
Keywords: Google fix, Gemini model, diversity controversy
News Content:
Google has recently announced a suspension of its Gemini model’s text-to-image function following revelations of a technical issue that刻意 Pursues “diversity,” as reported by BBC and The Verge. The problem lies in Gemini’s tendency to overemphasize gender and racial diversity in generated images, to the point of inaccuracies and historical inaccuracy.
User feedback suggests that Gemini leans towards depicting a wide range of genders and ethnicities in its images, sparking discussions on image authenticity and algorithmic bias. In response, Google has expressed its commitment to addressing these concerns and promptly taking action to eliminate potential biases, ensuring fairness and accuracy in its AI technology.
A Google spokesperson stated that the company aims to rectify the issue promptly and relaunch the improved Gemini image generation feature. This move underscores Google’s dedication to removing technological biases and maintaining user trust. As AI becomes increasingly prevalent in daily life, ensuring algorithmic fairness has emerged as a critical issue for the industry. Google’s response sets a positive example for the sector in tackling this challenge.
【来源】https://www.ithome.com/0/751/457.htm
Views: 1