多年来，我一直工作在一线课堂。相比单纯追求解题速度和考试技巧，我更关心一个问题：学生究竟是怎样真正理解一个数学概念的？
在教学中，我尤其重视概念的生成过程、数学直觉的建立，以及学生独立思考能力的发展。我希望学生不仅知道「这道题怎么做」，还能够理解：为什么会这样？这个概念为什么要被发明出来？如果没有它，我们会遇到什么困难？
因此，我的课堂通常不会从公式开始，而会从问题、矛盾、直觉和真实情境出发，让学生逐渐发现数学结构，再进入形式化表达和问题解决。
长期的一线教学也促使我持续研究数学教育。我逐步形成了一套以认知生成、主体性和深度理解为核心的教学思想，并将这些研究应用于数学课堂设计、课程开发和教育智能体的建设之中。
近年来，我开始重点研究人工智能与教育的结合，并开发教育智能体。我希望 AI 不只是替学生给出答案，而是能够理解学生当前的认知状态，判断他「卡在哪里」，再通过适当的问题、解释和反馈，引导学生继续思考。对我而言，AI 教育真正值得研究的问题不是「AI 能不能把题做出来？」，而是——
「AI 能不能帮助一个孩子学会思考？」
我相信，好的数学教育最终留下的，不应该只是学生会做多少道题，而是他面对未知问题时，仍然愿意问一句：为什么？并且知道怎样继续往下想。
—— 李 昕
For many years I have worked in the classroom. Rather than chase speed of solution and examination technique, I care about one question above all: how does a student actually come to understand a mathematical concept?
In teaching I give particular weight to how a concept is generated, to the building of mathematical intuition, and to the growth of a student's capacity to think independently. I want a student to know not only how this problem is done, but also why it is so; why this concept had to be invented at all; and what difficulty we would run into without it.
My lessons therefore rarely begin with a formula. They begin with a problem, a contradiction, an intuition, a real situation — letting students find the mathematical structure for themselves before moving on to formal expression and problem solving.
Years in the classroom have also kept me researching mathematics education. Gradually I have formed a body of teaching thought centred on cognitive genesis, on the student as a subject, and on deep understanding, and I apply it in lesson design, curriculum development and the building of educational agents.
In recent years my main focus has been the meeting of artificial intelligence and education. I would rather AI did not simply hand a student the answer, but understood where that student's thinking currently stands, worked out where he is stuck, and then — through the right question, explanation or piece of feedback — led him to carry on thinking. To me the question worth researching is not whether AI can solve the problem, but —
“Can AI help a child learn to think?”
What good mathematics education finally leaves behind, I believe, should not be the number of problems a student can do. It should be that, facing a problem he has never seen, he is still willing to ask: why? — and knows how to go on from there.
— LI XIN
