The Duality of AI in Software Engineering

What We’re Gaining, What We’re Losing, and Why Simplicity Still Matters

We’ve spent decades trying to speed up software delivery. Better tools, better processes, better-sounding architectures: each wave has promised to streamline how we build software. Yet some have actually added complexity, and in doing so, slowed us down.

Now, AI coding assistants seem like they could finally deliver on that promise. They offer reduced boilerplate, faster workflows, and more time for meaningful work. The reported productivity gains are impressively - perhaps even irresistibly - positive.

But this isn’t just another tool. It’s a fundamental shift in how we build software, and recent research is starting to expose less desirable second-order effects: more code, more duplication, and a level of hidden complexity that’s harder to see and even harder to unwind.

This talk explores the tension between speed and simplicity. I’ll share what we’re beginning to learn about the long-term impact of AI-assisted development, why keeping systems (and organisations) simple still matters, and what leaders should be paying attention to before short-term gains turn into long-term costs.