Ever notice that your teammates that are all in on ChatGPT, Cursor, and Claude for their development projects are far from being your strongest teammates? They scrape by at the last minute to get something together and struggle to ship it, and even then there are glaring errors in their codebase? And meanwhile the strongest developers on your team only occasionally run a prompt or two to get through a creative block, but almost never mention it, and rarely see it as a silver bullet whatsoever? I have a theory that a lot of the noise we hear about x% (30% being the most recent MSFT stat) of code already being AI-written, is actually coming from the wrong end of the organization, and the folks that prevail will actually be the non-AI-reliant developers that simply have really strong DSA fundamentals, good architecture principles, a reasonable amount of experience building production-ready services, and know how to reason their way through a complex problem independently.
[link] [comments]