There’s a strange split right now.
On one side, people are saying AI is breaking everything — unreliable, unsafe, causing issues in production.
On the other side, people are claiming you don’t even need engineers anymore.
Truth is… both sides are reacting to something real, just from different angles.
What actually changed
For years, a big chunk of software work was repetition.
Building APIs, dashboards, integrations - not trivial, but also not deeply complex most of the time. You could get very far just by being fast and experienced.
And that was enough.
Then AI came in and got really good at exactly that layer.
Now the same work:
- can take minutes instead of hours
- can be generated from a rough prompt
- looks “good enough” most of the time
That’s where the shift happened.
What actually got replaced
Not engineering.
But that middle layer of:
knowing just enough to build things, but not enough to fully understand them
That layer got squeezed hard.
Because if your main value was speed or execution, AI now does that faster.
The uncomfortable truth is that makes most of Software Developers worldwide.
The part people underestimate
AI is fast.
But it’s not accountable.
It doesn’t:
- think through edge cases properly
- understand long-term system behavior
- care about maintainability
- own the outcome
So yes, you can ship faster.
But also… things break in ways that are harder to debug.
And that’s already starting to show in real systems.
The gap is getting bigger, not smaller
There’s this idea that AI is leveling the field.
Honestly, it’s doing the opposite.
Good engineers are now:
- moving 5–10x faster
- using AI as leverage
- catching bad outputs early
Average engineers:
- rely too much on generated code
- ship things that “seem fine”
- struggle when something goes slightly off
So the difference between the two is actually increasing.
What still matters (more than before)
Speed used to be a differentiator.
Now everyone is fast (but not careful).
What matters now:
- understanding the problem deeply
- designing systems properly
- knowing what to trust and what not to
- communicating clearly
These were always important, but now they’re the whole game.
Where this is going
AI didn’t remove the need for engineers.
It removed the comfort zone where you could operate without deep understanding.
And that’s why there’s so much confusion right now.
Some people are seeing AI replace the work they used to do.
Others are using AI to become significantly better.
Both are true.
We’re still early, and honestly still figuring this out.
But one thing is clear:
You can build faster than ever.
But if you don’t understand what you’re building,
you’ll also break things faster than ever.
One thing is for sure, software engineering, the way it used to be, will never be the same again.
Adapters will be the winners.