LLMs Will Not Replace Software Engineers

With the rise of large language models (LLMs) like GPT-4 and Claude 3 Opus, some people are wondering if software developers will soon be out their jobs.

Here's why LLMs won't be replacing software engineers:

1. The big picture: LLMs can't architect software or plan projects to match business requirements. They're just not good at it. You still need experienced senior devs for this. (From my understanding, this is fundamentally due to the way LLMs work and thus won't change.)
2. Prompt engineering: To get good code from an LLM, you need to tell it exactly what to do. The quality of the prompt determines the quality of the output. And to write a good prompt, you need to anticipate what the code will look like.
3. Bug fixing: LLM-generated code often has subtle bugs or issues. You need to be able to spot and fix these quickly. If it takes you longer to fix the generated code than to write it yourself, what's the point?

The better LLMs get at generating code, the better software engineers need to be to use them effectively. LLMs are tools, not replacements.

If anything, they'll make the job more interesting and challenging.

(Some people worry that junior engineers won't be able to become seniors in a world with LLMs. But people are smarter than that. Junior devs who grow up with LLMs may even have an advantage in learning how to use them effectively.)

What I'm building

Delegate tasks. Get software.

Give Vroni a GitHub issue, bug report, spec, or rough idea. It reads the repo, plans the change, writes code, runs checks, and works toward a review-ready pull request.

Take a look at vroni.com