Is AI Coming for Software Engineers?

Image generated with ChatGPT

There’s been a lot of talk about a future where AI takes everybody’s job. Even software engineers are feeling anxious after seeing demos of Devin, an AI SWE, building apps and fixing errors autonomously.

But will these AI become so good they can manage and drive projects end-to-end?

I think AI will develop more as assistants to human software engineers, rather than replacing them. AI will take care of boring stuff like coding and fixing bugs, and let humans focus on driving technical direction and working on domain-specific tasks. After all, software engineering isn’t just about writing code.

I think there will be less demand for entry-level software engineering jobs because they mainly focus on execution. But senior+ engineers are safe because they focus more on dealing with incomplete requirements and decision-making, solving domain-specific problems, mentoring teams, and cross-functional collaboration — tasks that AI can’t handle.

A role of a software engineer will evolve though — a job will require knowing how to integrate AI features into systems, mastering AI-powered tools and IDEs, investing more in human-only skills, like the ones I mentioned above.

For anyone wondering how to adapt to these changes, here are my suggestions:

1. Invest in “human-only” skills like clear and effective communication, product vision (focusing on what to build, not how), and dealing with incomplete or conflicting requirements.
2. Get comfortable with AI-powered IDEs. You’ll see how fast it is to bootstrap a project and understand that coding is just one part of the job.
3. Master prompt engineering — it will enable you to ship code faster, write design documents fast, and much more.

In my view, a jobless future for software engineers is still far away. What I believe is that the future of software engineering won’t be one where human engineers are replaced, but one where they and AI work together to push the boundaries of what’s possible.