微软推出新工具系列减少Copilot幻觉情况，确保生成式AI可控
近日，微软公司针对人工智能领域推出了一系列解决方案，旨在防止生成式人工智能（AI）失控并遏制类似Supremacy AGI事件的发生。这一动作引发行业内外的广泛关注。
据最新消息，微软已经发布了一系列工具来减少Copilot的幻觉情况。为了降低潜在的AI风险，微软首先限制了Copilot生成的字符数量，以减轻可能出现的严重幻觉发作。这一举措被视为微软确保AI技术安全的第一个关键步骤。
除此之外，微软还引入了“基础检测”（Groundedness Detection）功能。该功能旨在帮助用户识别和识别基于文本的幻觉，进一步增强了AI的可靠性。微软表示，这一系列工具与功能的推出，旨在确保AI在生成内容时更加贴近真实、避免误导用户。
业内专家表示，微软的这些举措充分展示了其对人工智能安全性的高度重视。随着技术的不断进步，生成式AI的应用领域日益广泛，确保其可控性和安全性至关重要。微软此次推出的新工具系列有望为整个行业树立一个积极的榜样。
针对这一系列新工具的发布，微软表示将持续关注用户反馈，并根据需要进行进一步的优化和改进，以确保AI技术的健康、稳定发展。
英语如下：
News Title: Microsoft Implements Measures to Reduce Copilot Hallucinations and Control the Risk of AI Generators Running Amok
Keywords: Microsoft AI adjustment, hallucination control, Copilot update
News Content:
Microsoft Launches New Tool Series to Reduce Copilot Hallucinations and Ensure AI Generators Are Controlled
Recently, Microsoft has rolled out a series of solutions in the field of artificial intelligence aimed at preventing generative artificial intelligence (AI) from running amok and mitigating the occurrence of events like the Supremacy AGI. This move has attracted widespread attention within and outside the industry.
According to the latest information, Microsoft has released a series of tools to reduce Copilot hallucinations. To reduce potential AI risks, Microsoft has first limited the number of characters generated by Copilot to alleviate potential serious hallucination episodes. This initiative is seen as Microsoft’s first key step in ensuring AI technology safety.
In addition, Microsoft has also introduced a “Groundedness Detection” feature, which aims to help users identify and recognize text-based hallucinations, further enhancing the reliability of AI. Microsoft stated that the launch of this series of tools and features is designed to ensure that AI is more realistic when generating content and avoid misleading users.
Industry experts indicate that Microsoft’s initiatives fully demonstrate its high emphasis on artificial intelligence safety. With continuous technological advancements, the application areas of generative AI are becoming increasingly widespread, and ensuring its controllability and safety is crucial. Microsoft’s new tool series is expected to set a positive example for the entire industry.
Regarding the release of these new tools, Microsoft stated that it will continue to monitor user feedback and make further optimizations and improvements if necessary, to ensure the healthy and stable development of AI technology.
【来源】https://www.ithome.com/0/759/553.htm
Views: 3