Junior → Senior in the Age of AI series · Chapter 4 of 6
The previous chapter ended with two questions: if the Mid-level absorbs the work from below and from above, what is left for the Senior? And if nobody trains Juniors, where do the next Seniors come from?
In the first three chapters the Senior always lands on the winning side. Their employment stays stable, job openings look for them and AI makes them produce more. They seem to be the only level that can watch all of this calmly. The data suggests that calm comes with a catch.
The one who thinks they are faster
In 2025, the research center METR ran a controlled experiment with 16 developers with years of experience, working on projects they knew inside out. Some tasks allowed AI and others did not, assigned at random. With AI, they took 19% longer.
The most striking part came afterwards. Before starting, they expected AI to make them 24% faster. And when they finished, even after taking longer, they still believed AI had saved them about 20% of their time. The extra time went into writing instructions, waiting for answers and reviewing what AI proposed. They accepted less than half of what it generated.
The study has clear limits: a small group, one specific type of work and tools from early 2025. But it leaves a warning that applies to any field: the feeling of speed and actual speed can go in different directions, especially for someone who already masters their work.
The reviewer of everything
There is a second, quieter effect. When everyone produces more with AI, someone has to review that volume, and it is almost always the Senior.
A 2026 Harness survey of 700 engineering professionals in five countries found that 81% spend more time reviewing work since their team adopted AI tools. What used to be a quick review of a colleague's work is now reconstructing what a machine meant to do from a poorly written request.
The Senior went from being the one who solved the hard problems to being the quality control for everything that ships. Nobody chose them for it. They were simply the only one with enough judgment to do it.
The question nobody wants to ask
In a McKinsey analysis, two engineering leaders from a large tech company summed it up this way: AI gives the Senior a boost that multiplies their output, while it puts a brake on the person starting out, who does not yet have the judgment to steer and verify what AI delivers. The resulting incentive is to hire Seniors and automate Juniors. And that slowly takes apart the base of the pyramid every Senior comes from.
Companies know it. In an IDC survey for Deel, almost seven in ten said there are now fewer on-the-job learning opportunities for Juniors, and 71% reported more difficulty recruiting and training their future leaders.
Today's Seniors learned by getting things wrong in tasks AI now does, with someone next to them explaining why. If those tasks no longer exist and that someone is busy reviewing what AI generates, the cycle breaks.
What the Senior can let go of and what they cannot
| What they can let go of | What they cannot let go of |
|---|---|
| Writing everything themselves | Deciding what gets done and what does not |
| Being the fastest on the team | Being the one who catches the error nobody sees |
| Reviewing line by line | Defining how review works and what is accepted |
| Keeping knowledge in their head | Passing it on to someone else |
In practice:
- Measure before believing. If AI really saves you time, it should show in what you deliver, not just in how you feel.
- Turn review into written criteria. The rules you apply when reviewing, written down, let others review without depending on you.
- Train on purpose. Training no longer happens on its own in the daily work. It needs reserved time, real tasks for the person starting out and room to make mistakes where mistakes are cheap.
- Explain the why, not just the fix. A correction without explanation fixes a deliverable. With explanation, it trains a person.
The Senior who comes next
Seniors are not going to disappear. What is at risk is the next generation reaching that level. And that depends, to a large extent, on what today's Seniors do with the time AI frees up for them, if it really frees any up.
If you are a Senior and feel you spend the day reviewing instead of building, or you lead a team and are not sure how to set up a plan for your Seniors to train the people coming next, send me a message and let's talk.
Next chapter
If all this is so obvious, why do so many teams hold AI back, or use it in secret? In chapter 5 I look at resistance to change, which often does not come from where you would expect.
The full series
- If nobody hires Juniors, where will the Seniors come from?
- How does a Junior gain experience if AI does their job?
- The Mid-level now does the work of three. On the salary of one
- The Senior with AI thinks they are faster. The data says otherwise (you are here)
- The four faces of resistance to AI
- In today's job interview everyone has AI. The winner is whoever can explain what they did
Sources
- METR's study on how AI affects developer productivity, DX
- AI coding tools could make experienced software engineers less productive, Business Insider
- The review tax: why 81% of developers are buried in AI code review, DEV Community, on the Harness 2026 report
- Building expertise in the age of AI: who trains the next generation, McKinsey Quarterly
- Enterprises are cutting back on entry-level roles for AI, ITPro, on the IDC study for Deel




