I think software engineers will still exist. Software engineering certainly will.

1. LLMs make “Don’t do the same thing twice” easier.

If your job depends on repeatedly rebuilding solutions that already exist, often badly, LLMs are a threat. Great engineers already follow rigorous principles and best practices to eliminate repetition, innovate, and push software’s boundaries.

LLMs also lower the programming bar. Some people build software without even thinking of themselves as programmers. Others are simply users of a powerful piece of software. Natural language can become a programming interface; the distinction depends on how you use it.

People without a programming background can become good engineers by learning those principles. As the language barrier falls, the principles become more valuable.

2. Less repetition improves efficiency. Innovation drives business outcomes.

Absorbing human knowledge and applying it to tasks is the model’s greatest strength. The model itself is a remarkable piece of software. I think of it as a knowledge recommender: it learns from what people have already figured out and makes that knowledge useful in response to your query.

That explains why I see its biggest contribution so far in efficiency. It lets us stop repeating work that someone else has already done.