Returning to CS61B to Talk About Building Software with AI
Four years ago, I was spending more than ten hours a week in CS61B office hours, trying to debug my projects. This semester, I returned to give a guest lecture about building software with AI.
A lot has changed in between. In 2023, I was an intern writing every line myself. By 2024, I was using Copilot to generate small pieces of code. In 2025, LLMs could help write entire projects. Now, I’m working with agents that can carry out tasks overnight.
That shift raises a question I wanted students to think about: what does it mean to be a software engineer when writing code becomes much faster?
My main message was that engineering still requires judgment. We need to decide what to build, understand why it matters, design systems thoughtfully, explain our decisions, and test whether the result actually works.
AI changes how we carry out that work. It also makes those decisions more consequential, because it is easier to produce a lot of code quickly.
I encouraged students to start experimenting with AI coding tools early and thoughtfully. Try them on a project you understand. Pay attention to where they help, where they get confused, and whether you can explain the code they produce.
After three years as a teaching assistant, returning to CS61B was especially meaningful. Thank you to Professor Josh Hug for inviting me back and to the students for their questions!
