AI didn't erase the junior engineer's value, it increased it it

AI didn't erase the junior's value. It accrued it.

August 18, 2026 · 4 min read

“AI ate the junior’s marginal value”. This is the feedback I sewage from the past station connected this topic. According to this argument, the activity of a inferior technologist goes thing for illustration this:

• There is simply a request for a solution. Someone much knowledgeable (a elder engineer) plans it and designs it.
• The elder technologist breaks down the activity successful aggregate mini tasks, clarifying the steps to beryllium taken. Each task is fixed to a inferior engineer.
• The inferior technologist executes it, which nowadays intends prompting it to an AI tool, and creating a propulsion petition (PR).
• The PR receives feedback from much elder engineers. The inferior technologist gets the feedback and takes it to the AI instrumentality again, proposing changes.
• Repeat until done.

If the process supra is what happens successful practice, why do we request a inferior engineer? They are conscionable passing done requests (and sometimes adding noise) from 1 spot to the other, and they costs a afloat clip salary. That’s fair.

Before we jump to conclusions, let’s look astatine a different story. Or possibly the aforesaid story, approached successful a different manner.

In our product, location was a characteristic which had been requested for years, but had not been built yet. It wasn’t overly complex, but it was not critical. Or astatine slightest not captious for capable customers to make it done the prioritization threshold. This meant that twelvemonth complete twelvemonth immoderate customers were disappointment they didn’t person a characteristic they needed.

This summer, we assigned the problem to an intern (that’s little tenure than a inferior engineer). The intern led the improvement of this feature. They talked to the merchandise head to understand the problem and requirements. They wrote the creation archive connected really to attack it, aligned pinch the team, and built it. Of people they did that pinch the thief of AI, and the squad they were moving with.

But they led it. They dealt pinch the inconsistencies, pinch knowing trade-offs, some successful the method and merchandise areas. They adapted the attack arsenic they recovered issues. And delivered value. AI did nutrient overmuch of the code, and they owned the decisions. A customer problem that would not person been solved has now been solved. By an intern, astatine a very debased costs for the company.

All engineers negociate complexity. What changes is really much.

Beyond the truth that immoderate institution will request elder engineers successful the future, location are very pragmatic reasons why inferior engineers are still valuable.

Junior engineers still adhd capacity to an organization. And that’s because the activity of an technologist is not to constitute the codification (or punctual an AI tool) according to a spec. It is to lick a customer problem pinch software, managing the method complexity that exists successful deciding really to. That applies to each engineers. Staff engineers negociate a batch of complexity. Junior engineers negociate a small complexity. But the domiciled is the same.

This goes beyond coding. It requires knowing the problem, and besides the customer perspective. It requires knowing that if you build a characteristic successful a peculiar manner, it will lead to a peculiar group of trade-offs. And those trade-offs are beyond what AI tin decide, because they require discourse that is broader than the codebase. Technical determination making is still very important and inferior engineers adhd to it, enabling an statement to do more.

Training really became overmuch cheaper. Most of the training costs for an early tenure technologist is to understand the company’s method context, meaning the nuances of a ample codebase and architecture, and to supply basal training connected software, for illustration the language, the tooling, the patterns.

In the past, this either required meaningful proactivity from a inferior engineer, researching topics and studying, aliases it required existent quality effort from much elder peers, who would walk clip explaining basal concepts. This effort has not been eliminated, arsenic discourse coming from humans is still cardinal for productivity successful software. But a batch of this activity tin beryllium short-circuited pinch the usage of AI.

Being AI-native is valuable. We cannot, arsenic an industry, opportunity that engineers request to beryllium AI-native successful occupation descriptions and past not prosecute the group that fresh this explanation the best. If the presumption is that AI is going to radically simplify the method information of the role, past the group who person started their careers pinch AI will beryllium successful the champion spot erstwhile they person acquired the experience.

Going backmost to the earlier example, the characteristic the intern shipped had been requested for a agelong time. It wasn’t important capable to get prioritized, but required excessively overmuch judgement to manus it to AI. These tasks beryllium and will proceed to beryllium successful each team. AI expands what each level tin handle, including juniors.

More important than that, engineering organizations will proceed to request method judgment. And the early judgement for your statement should beryllium increasing correct now.