Every other headline wants to tell you that AI is coming for your job. It makes for a great panic, but it’s the wrong frame. AI isn’t going to make engineers redundant. It’s going to make a handful of things we’ve spent our whole careers wrestling with redundant – and honestly, good riddance to most of them.

Here are the four I’m betting on.

1. Technical debt stops being a problem

Technical debt doesn’t disappear. It just stops mattering the way it used to.

The reason debt has always been crippling is that paying it down is slow, manual, and risky. You inherit a sprawling codebase, you spend weeks just building a mental model of it, and by the time you understand it well enough to safely change it, the change itself is the easy part. The cost was never the edit – it was always the comprehension.

AI collapses that cost. It can synthesise an entire repository faster than any human can read the README, trace the dependencies, and work out what actually needs to change to ship a feature. Rewriting a gnarly module that everyone’s been too scared to touch for three years stops being a quarter-long project and becomes an afternoon. The debt still exists in some accounting sense, but when the interest payment drops to near zero, it’s no longer the thing dictating your roadmap. The scary legacy system loses its teeth.

2. Boilerplate dies, and templates go with it

Be honest: most of us have written code whose entire job was to write other code. The scaffolding generators. The “new project” templates. The config that exists only so the next config can exist. It’s a strange little genre of engineering where one builds the mold so they can pour the same shape a hundred more times.

That whole layer evaporates. You don’t write a template when you can describe the outcome and have the structure generated for you, correctly, every time, adapted to the specifics instead of forced into a generic shell. We won’t maintain boilerplate because there won’t be any to maintain. The repetitive substrate of software – which is the part nobody ever bragged about writing – is exactly the part AI is best at making disappear.

3. Languages stop being a moat

Knowing a language deeply used to be a credential. You were a “Rust person” or a “Go person,” and switching cost you real time and real fluency.

That distinction is fading fast. What still matters – and what will always matter – is the conceptual layer: algorithms, control flow, data structures, knowing why a loop is the wrong tool and a hash map is the right one. That’s the actual engineering. The syntactic specifics of any given language? Increasingly a lookup, not a skill. You can hand AI an unfamiliar codebase and get it explained back in plain English, then have your intent translated into idiomatic code in a language you’ve never formally learned.

Languages don’t vanish, but they stop being walls between you and a problem. They become interchangeable surfaces over the same ideas – and the engineers who understood the ideas all along were never really betting on the syntax anyway.

4. Clever code stops earning applause

This is the one I think we’ll mourn a little, even though we shouldn’t.

There’s a holy grail in engineering culture: the syntactically sweet solution. The dense, elegant, almost showing-off bit of code that makes a senior dev lean back and go “…nice.” We’ve revered the artistry of it, the cleverness, the compression, the craft visible in the source itself.

That reverence is on its way out. When AI can produce a working, performant solution on demand, the how of the code – as in how creatively you arranged it, how impressively terse it is – stops being the point. The conversation moves entirely to outcomes. What does this do? Is it correct? Is it fast enough? Nobody’s handing out medals for the cleverness of the implementation anymore.

It’s the same arc as the trades. A brilliant electrician or plumber does work of real skill and beauty – and then it’s sealed behind a wall, never to be admired, judged only by whether the lights come on and the water runs. Software is heading behind the same wall. The craft is still real. It’s just no longer on display.

So, who actually matters?

If standard software development is getting commoditised – and it is – the obvious question is who’s left standing.

The answer is the curious ones.

When the mechanical parts of the job collapse, productivity doesn’t shrink, it explodes. One engineer can now reach across domains, languages, and systems that would have taken a whole team a year. But that leverage only goes to people who can actually wield it: people who can context-shift quickly, who want to understand why something works rather than just that it does, who treat a new domain as a thing to be devoured rather than a thing to be feared.

The engineers who defined themselves by mastery of a language, by their boilerplate, by their clever one-liners – well, they were defined by exactly the things going redundant. The engineers who defined themselves by curiosity were betting on the only thing that survives.

AI isn’t coming for the engineer. It’s coming for the parts of engineering that were never really about thinking. What’s left is the most human part of the whole craft: wanting to know how things work and being delighted to find out.

Stay curious. That was always the job.