The Role Collapse Cycle in Tech

~600 Words | ~2.5min Read

I’ve been giving talks on AI and software delivery lately. Same questions keep coming up. “What happens to QA roles?” “Will developers still be needed?” “Should I be worried about my job?” The anxiety is real. But in my research, I found a something: We’ve been here before.

When telephones first emerged, we had operators. When computers first emerged, we had computer operators. But today, who thinks to call someone else to open your email for you!?

Programmers didn’t stop existing when C++ came out (Or Python, or Javascript or…). QA didn’t ‘disappear’ when Selenium came on the scene either. But the nature of their job changed. Programming moved from moving bits between registers, into objects, methods, and variables. QA moved from test steps, into test scripts, and testing harnesses. Each time, the abstraction stack moved up. The specialized roles consolidated, or took on new skills. This isn’t unprecedented disruption. AI is something like the fifth iteration of a recurring pattern.

AI replaces tasks within jobs, not entire jobs. A developer doesn’t stop being valuable because AI can write code. But the value a developer adds does shift. We move from doing the work to creating systems that do the work. That’s a different skill. But it is learnable. The people best positioned to learn it are already on your teams.

This is a double-loop learning moment. We’re not asking “how do we do today’s work better?” We’re asking “should we be doing this work at all?” Before we can optimize execution, we need to question direction. What contributions actually matter? Which do we actually want to pay for?

The answer isn’t throwing people out. It is helping them convert existing knowledge into new skill forms. Quality didn’t stop being valuable! But the kind of work to help create quality changed. Your best QAs already think about quality systems, not just testing. Your best Scrum Masters already think like process engineers. These aren’t new people you need to hire. They’re capabilities you need to develop in the people you already have.

In the Effective Executive, Drucker proposes a key question for all knowledge workers. “What can I contribute?” That question matters even more now. Our roles will change. They always do. The cheese will move again. We just need to learn to find new cheese.

Think musket fire before launching cannon volleys. Start small. Define what success looks like in one role before restructuring your entire org. Train one person in the new skill before rolling it out to the team.

This is how previous transitions succeeded. When Test Automation emerged, the best companies didn’t fire their QAs. They trained them to think about like programmers alongside test design. When DevOps emerged, the successful organizations didn’t hire entirely new teams. They taught developers operations thinking and operations teams development practices. When Node.js unified the stack, the best full-stack developers weren’t fresh graduates. They were frontend or backend specialists who learned to combine their existing expertise with the new tooling.

The pattern is clear: skill stacking beats replacement. Combine your existing domain knowledge with the new abstractions. Quality must be designed into your new AI-assisted SDLC. AI Systems that call out for human judgement when it’s needed. Process flow build into the fabric of the work.

The best way to build for tomorrow is to properly define the new necessary contributions. Train the players who are already serving in some other role on your teams. We’re creating a new way of working. The people with the best insight to help us are already here. We just need to invest in them.