构建以人为本的课堂AI(并让学生参与塑造)

📄 中文摘要

多数AI演示通常光鲜亮丽,但很少经过目标用户的实际测试。一个课堂AI项目通过优先进行“不适”的部分,即将其置于真实学生面前并仔细倾听反馈,从而启动。该项目并非始于商业计划或宏大承诺,而是源于一个简单问题:AI能否真正帮助学生更好地学习,而不仅仅是更快地回答问题?为此,团队构建了一个小型课堂演示并早期分享。收到的并非炒作,而是真实的反馈。一些学生喜欢提问的便捷性,但也有学生指出了实际问题,例如答案重复而非适应性变化,以及主题未能有效分离。

📄 English Summary

We Are Building a Human-Centric Classroom AI (And Letting Students Shape It)

Most AI demonstrations appear polished, yet few undergo testing by their intended users. A classroom AI project began by tackling the uncomfortable part first: exposing it to real students and meticulously listening to what broke. This initiative didn't originate from a pitch deck or grand promises, but from a fundamental question: Can AI genuinely enhance student learning, rather than merely accelerating answer delivery? Consequently, a small classroom demo was built and shared early. The feedback received was not hype, but authentic insights. While some students appreciated the ease of asking questions, others highlighted significant issues. These problems included answers repeating instead of adapting to context, and topics failing to separate clearly, indicating a need for more sophisticated conversational flow and content organization. This human-centric approach prioritizes user experience and iterative development based on direct student input, aiming to create an AI tool that truly serves educational needs by addressing practical shortcomings identified during early testing.

Powered by Cloudflare Workers + Payload CMS + Claude 3.5

数据源: OpenAI, Google AI, DeepMind, AWS ML Blog, HuggingFace 等