im graduating in SWE soon but Claude does all my thinking. am I actually learning?

im going into my senior year as a SWE major and honestly im starting to panic about my actual baseline competence. Our curriculum is heavily Java and Spring Boot. a year or two ago, if I got stuck on a project, I’d actually break down the problem, read docs, struggle with stack traces, and eventually figure it out. Now? my default reaction to literally any roadblock is to alt-tab and open Claude. it started innocently enough. I'd paste an error log and ask it to explain a weird JVM exception, or have it write a quick regex. Then it crept up to 'refactor this controller.' Now, its basically full-scale delivery. I outline the requirements, let it plan the architecture, and it just writes the implementation. The gap between 'I can make this run' and 'I actually understand how this runs' is getting dangerous. The tooling right now is just too good at wiping out the friction you normally need to actually learn anything. Between using Claude Code to let agents operate directly on my repo, or using tools like Lovable, v0, or Enter Pro to just spit out whole web apps with the db and auth already handled (which is crazy fast), the barrier to shipping is basically zero. I put together a course project last week that works perfectly. But if a professor or an interviewer asked me to whiteboard how the Spring DI container is managing my beans in that project, or to trace exactly where a database connection pool is hanging up without internet access? I'd propably blank. im not anti-AI, and im definitely not going to stop using Claude. the speed is just too insane to ignore, and I know this is how the industry works now. But I feel like I'm accumulating massive cognitive debt. I'm basically outsourcing the actual learning process to the model. If you've been vibe coding or relying heavily on Claude Code workflows, how do you manage this? Specifically: Where do you draw the line between 'I need to write/debug this myself' and 'I'll just let Claude handle it'? How do you force yourself to do line-by-line reviews when the code already runs? I try, but I get lazy instantly. Outside of interviews, how are you testing your own raw debugging skills to make sure they haven't completely atrophied? I seriously need to fix my workflow before graduation because right now I feel like an imposter.

添加评论
点赞收藏
点踩分享查看原文
评论
?
参与讨论