The Semantic Telemetry Project focuses on understanding human-AI interactions in Microsoft Copilot through a novel data science method. This is essential for recognizing how users apply AI to real-world tasks, offering insights to enhance use cases and identify improvement areas. Recently, the project shared its approach of classifying chat log data using large language models (LLMs) to analyze interactions efficiently and in near real-time. Two LLM-generated classifiers, Topics and Task Complexity, were introduced. These classifiers act as early indicators of user engagement and reveal variations in usage and satisfaction based on AI and user expertise. Key findings indicate that users tackling professional, technical, or complex tasks are more likely to continue engaging with the system, emphasizing the importance of task nature in driving sustained interaction and satisfaction. The project aims to refine user experience by leveraging these insights into user behavior and task complexity.
本专栏通过快照技术转载,仅保留核心内容

内容中包含的图片若涉及版权问题,请及时与我们联系删除



评论
沙发等你来抢