StudyBlog

Thoughts on AI Coding

2025-07-14

What do you think about AI coding? It's going to spread across the industry before long, so here are my personal thoughts before that happens.

Strengths and Weaknesses

For incremental development on top of a well-defined framework, AI produces good quality code. IaC and SQL are also areas where AI is said to excel — and I fully agree.

On the other hand, system migrations or language conversions (e.g. COBOL to Java) seem quite difficult for AI. It might work if you define the framework first, write some features by hand to establish the rules, and then let AI follow those patterns for the rest.

Advice for Junior Engineers

I think junior engineers should deliberately set aside time to code without AI. In a company environment, speed is prioritized, which makes this hard — but I believe it's worth pushing through mentally.

To become an engineer who can truly leverage AI in the future, hands-on coding experience is essential.

― From a 5-year SE

← Back to list