Will Average Engineers Become Obsolete in the AI Era? The Day “Technical Skill” Changes Meaning

Technology companies in the United States and China are reducing their engineering headcounts.

One factor behind this is the increase in development productivity brought about by AI.

From a Japanese development team, this may still look like someone else’s problem. But when you actually develop software with AI, it becomes clear that this change is not temporary.

There are now situations where one or two people who know how to use AI can carry out development that would previously have required five.

The effect is particularly large in fields such as frontend development, where implementation patterns are relatively well established.

The work itself has not disappeared.

It simply no longer requires as many people to do the same work.

“Knowing This Technology Well” Is No Longer a Competitive Advantage

Until now, an engineer’s market value has depended largely on proficiency in particular technologies.

Deep knowledge of Java. Skill with React. Extensive experience building systems on AWS.

Engineers have been valued for understanding a particular language or framework deeply and writing code quickly.

But AI possesses an enormous amount of technical knowledge and can generate, revise, and test code.

Simply being able to “write React code quickly” is not as valuable as it once was.

What matters is the ability to decide what should be built, have AI implement it, and evaluate whether the resulting output is correct.

From the ability to write code to the ability to have code written and bring the product to completion.

The very nature of the technical ability expected of engineers is beginning to change.

Will the Survivors Be Elite Specialists or Full-Stack Engineers?

I believe the value of engineers will divide broadly into two directions.

The first is people with advanced expertise that even AI cannot easily replace.

This includes areas requiring deep experience and judgment, such as extremely difficult performance optimization, security, and specialized domain knowledge.

The second is full-stack engineers who understand the entire system and can drive development forward.

They understand the requirements, design the architecture, and maintain a view across the frontend, backend, and infrastructure.

The actual coding can be left to AI.

What matters is understanding how the parts connect and what could go wrong.

For example, it has traditionally been normal to assign separate people to the frontend, backend, and infrastructure.

But if AI assists with implementation in each area, the scope that one engineer can oversee expands dramatically.

In that environment, engineers with only moderate knowledge of one particular area will struggle.

Become an elite specialist, or learn to handle the entire system. The space available to engineers caught somewhere in the middle is steadily shrinking.

Recently, I Am Not Sure Whether I Am Still an Engineer

For me, using AI in development has recently become routine.

I organize the requirements, decide on the design approach, and give instructions to AI.

I review the code it produces and have it make corrections when there are problems.

Code that once took me hours to write is now completed in a short time.

It is a convenient era.

But a thought sometimes occurs to me.

Am I still working as an engineer, or have I become a director managing an exceptionally capable subordinate called AI?

I now spend more time instructing AI and checking its output than writing code myself.

There is no doubt that the center of my work has shifted.

Engineers were once craftspeople who built systems with their own hands.

Now the job is becoming more like assembling what AI has built and making the whole system work.

Technical Skill Is Not What You Know, but What You Can Complete

Knowledge of programming languages and frameworks will probably remain necessary.

But the era in which that knowledge alone determines an engineer’s value is coming to an end.

The question will be how complex a system you can complete with AI, and how few people you need to do it.

Companies will eventually realize this as well.

Some work can be done more efficiently by one engineer who knows how to use AI than by assembling several excellent engineers.

For engineers, this is both an opportunity and a threat.

As the definition of technical skill changes, so does the value of the experience we have accumulated.

In the past, engineers took pride in “code that only I can write.”

Now, “it gets completed without me writing it” is becoming a mark of an excellent engineer.

And eventually, an executive will probably ask:

“Then could AI replace that engineer too?”