AI Impact on Engineering

Engineers still have a place in the AI world. After the last few months in the AI boom, there has been a dramatic shift in how we work. People think that you can vibe code your way through every solution. It’s true that non-engineers can get a little taste of what it’s like to program something for fun. AI makes one-off projects fun and easy to accomplish and does a decent job at it. Long term, what engineers really need to do is learn the business they build solutions for. Something I have done throughout my career is make sure that I know the business need and problems inside and out. When I start development, I need to know the actual problem, not the acceptance criteria in a Jira ticket made two months ago. To be effective, we still need to understand the problem we need to work with or solve.

AI solves what we ask, not what we need

AI does very interesting things when we ask it to solve a problem. It creates solutions we really don’t need. Have you ever asked Claude Code to write you a line and it created a longer comment than the single line it wrote? Don’t get me wrong; I use AI every day when I work now, but it really hasn’t been around that long. People are needed just as much as they were before; we can just get things done a lot faster these days.

Speed without review

Speed to market has always been a problem for the powers that be. What people want is a solution, and they want it fast. When we start to throw AI at all of our problems, we just make more problems when we don’t review those solutions. When a business partner decides to vibe out a UI, it might look super cool and functional, but when you dig into their UI, it’s a solution that can’t scale. When that same business partner comes around in six months with a new idea, you might have just launched their other product which doesn’t fit cleanly with their new need. Engineers are still around for the purpose of vetting these solutions and architecting a solution around the problem. Most of my day is spent going back and forth with agents while I re-direct them and throw out made-up situations.

What I miss

I just don’t write much code anymore, if any. This is the saddest part of AI to me; I used to enjoy my long sessions and the sense of accomplishment when my solution worked. New engineers won’t be experiencing this, and it’s something that makes me sad. I want the engineers I’m mentoring to come in, ask me how to fix their code, what they can work on next, and get their hands dirty. When the engineers I mentor come to me now, it’s understanding how the business works and why solutions are being pitched to them in certain ways. This seems to be the path we’re going down, and it’s coming two to three years earlier in their career than I have seen before.

Where we go from here

I don’t think AI will be going away and we’ll continue to rely on it heavily. Engineers are needed in different ways now. Our roles have changed over the last few months: AI speeds up code but not solutioning. We’re left with more time to do things that we used to put on the back burner. We can sit with our business partners and still ship code; our late-night sessions might be going back and forth with people trying to solve architecture rather than code. I’m excited to see what else we’re able to do with our time and what everyone can create in our new world.