Will AI coding assistants, like GitHub Copilot, replace developers?

This was the burning question I explored during my presentation at the jPrime conference in Sofia, Bulgaria.

The verdict? While Gen AI won’t replace developers, it will transform our roles significantly. Future developers’ jobs will be less about coding and more about architecting innovative software solutions.

Key Takeaways

  1. Error-Proneness of AI: Given the error-proneness of AI coding assistants when generating new code or refactoring existing code, established software engineering practices like unit tests and code reviews are more relevant than ever. Only software development organizations that apply these practices can safely adopt AI coding assistants!
  2. AI for Code Understanding: All the hype around AI coding assistants is about how they make it easier to write code. However, software developers spend considerably more time reading code, especially for legacy applications. Therefore, using AI coding assistants to understand code can help us speed up software development.
  3. New Development Paradigm: Integrating large language models (LLMs) with conventional software applications is a new software development paradigm that enables us to build innovative solutions that were previously impossible. Start exploring this new paradigm today!