Will Software Engineering Die? My Honest Answer.
I've been asked this question in at least a dozen Slack threads, two LinkedIn comments, and one uncomfortably earnest coffee chat in the last six months. So let me give an actual answer.
No. Software engineering will not die.
But — and this is the part people skip — a specific *version* of software engineering is already dying. And if that's the version you're practicing, you should pay attention.
The version that's dying
The job that's going away is the one where you get paid primarily to:
- Google syntax and copy Stack Overflow answers
- Write boilerplate CRUD that follows an obvious pattern
- Translate a clearly-defined spec into code mechanically
- Fix bugs by trial and error without understanding the system
AI is genuinely good at all of these things. Not perfect. Good enough to make a solo developer much faster, and good enough to reduce how many junior engineers a team needs for routine work.
The version that's thriving
What AI cannot do — yet, and I think for a long time — is:
- Understand the actual business problem, not the stated one
- Know which of the ten technically correct solutions is the right one for this team
- Navigate organizational constraints and legacy decisions
- Make judgment calls with incomplete information
- Take ownership and accountability for outcomes
What this means practically
If I were starting out today, I would ruthlessly prioritise:
Systems thinking. How does this system behave under load? What breaks first? These require mental models that come from experience.
Communication. The ability to explain a technical decision to a non-technical stakeholder is worth more than it's ever been.
Taste. Knowing what good looks like. Recognising when AI output is subtly wrong. This is the meta-skill.
The honest bottom line
The question 'will SWE die' is the wrong question. The right question is: 'what kind of engineer do I want to be, and is that kind made more or less relevant by AI?'