Code review used to be a predictable bottleneck. A senior developer would go through a junior’s pull request, flag any issues, explain the reasoning, and then move on to other tasks.

But in an AI world this is not that simple anymore.

The amount of code grew exponentially

Where a senior developer might previously review and validate 15 lines of junior code a day, AI tools can push that number to 150 or more. But the working hours of the person reviewing and validating didn’t stretch at that pace.

On paper, that may sound like a win, and in many ways, it is. Velocity increases and junior developers can move faster with less “hand-holding”, so teams can address larger scopes without proportionally growing headcount.

The thing is…

If you’re responsible for 10 times as much code, your job changes completely. You’re no longer a developer first and foremost.

Sure, you can also use AI to help you review code and accelerate the process. Many developers argue that this is a lifesaver in many cases, but you will always have to spend time explaining what the other person missed, and as the output grows, this will take a lot more time than what you previously had to spend in a world without AI.

So, the keyboard time is reduced but the revision calls, the overseeing decisions and quality control multiply.

For some developers, that’s an exciting evolution but for a person who got into the game for the love of the code, that can mean a quiet identity crisis.

AI amplifies developers in both directions

Nothing can stop good developers anymore. They were good before, and now, with AI tools, they can produce a lot more in less time.

But weak ones are getting harder to spot (at least temporarily) and a developer who doesn’t fully understand what they’re building can now produce code that looks right, passes surface-level review, and makes it into production without anyone spotting it.

Let’s be honest: this could also happen before. The difference is that now, we’re talking on a completely different scale.

Spotting the programmers who are generating volume without understanding what they are producing creates a real problem for teams: how do you hold on to quality when the usual quality control measures are being compromised?

This question doesn’t have an easy answer, at least for now, and many companies will have to take a closer look at their processes and culture to be able to find a new way to work.

What this means for senior developers

As a senior, your role is shifting and critical thinking matters even more now. The developer who understands why a solution works (and when it won’t) becomes even more valuable. So, if you don’t want to get left behind, here are some non-negotiables:

• Your review process needs to evolve. Reviewing AI-assisted code is a different skill from reviewing human-written code. The errors and patterns are different and the gaps are in different places. Developing that AI quality control eye can take time and practice.
• If your job used to include teaching juniors through code review, that teaching needs to focus on the thinking aspect instead of the practical one. Less “this line is wrong” and more “do you understand what this block is actually doing and why.”
• Management skills are no longer optional. If you’re overseeing more output, you need to be better at prioritisation, delegation and knowing what deserves deep scrutiny and what doesn’t.

IT teams need to evolve

AI has raised both the floor and the ceiling at the same time but not equally.

The best developers are becoming exceptional. They use AI to move faster while their fundamentals keep them grounded, catching what the model misses, asking better questions and building things that actually hold up.

The weakest developers are becoming a harder problem, exactly because they produce more, not less.

The teams and the leaders who figure out how to use AI’s productivity gains without losing sight of who actually understands the work are the ones who will build things worth building.