计算与应用讨论班

报告题目：How AI changed the way of coding, and what math / CS students can learn from it

报告人：李绳（Nova Vision）

时间：2026年04月15日（星期三），10:30

地点：海纳苑2幢202

摘要：If we compare the evolution of AI coding tools to changes in payment methods, my workflow over the past year has moved through three stages: “cash” (Q&A and copy‑paste), “credit card” (human–AI collaboration), and “mobile payment” (agents that execute). The real shift is not the tools themselves, but the cost of a single iteration. Software work still follows the same loop—write, measure, keep or discard—but AI makes that loop run much faster.

This talk draws on my observations from June 2025 to April 2026 across 10 personal projects and 10,166 commits to answer a concrete question: what has AI actually changed? My conclusion is not that AI replaces programmers, but that it reshapes how work is organized. AI started as an answering assistant, then moved into real workflows—editing across files, running commands, and supporting refactors. More recently, agents can plan, execute, test, and iterate toward a goal, while humans shift from “doing everything” to “setting goals, defining standards, and supervising.”

For math and CS students, the key is not model rankings or subscription tiers, but which practices can be adopted immediately. AI is lowering iteration costs, enabling faster learning, experimentation, debugging, and building. The question becomes: can you use it to run your own thinking-and-testing loop faster, more precisely, and further?

联系人：王何宇（wangheyu@zju.edu.cn）