Right now AI writes more and more code. Because of this, using statically typed languages is even more sane. Languages like Java and Rust give clear rules and strong checks.
I also prefer frameworks that work on a higher level. For example, 20251123122633⁝ SpringBoot 4.0.0 gives many features out of the box: better HTTP client, better observability, cleaner modules. This reduces low-level code and lets us focus on business logic. I think frameworks will keep going in this direction. They should also be easy for AI to use, but I am not fully sure yet what “AI-friendly Framework” really means.
For my own work, I feel that I only need only two languages:
- Java - for business systems, backend logic, and large projects.
- Rust - for very safe, system-level or performance-critical parts, and for fun projects.
Maybe
- TypeScript / javascript - for everything in the frontend. But to be completely honest, I feel like frontend part could be almost completely offloaded to AI. Or even the simple usage of higher level frameworks like https://nextjs.org/ etc .. Ai here can scaffold something useful much better then me.
Why not python:
- Can we refactor safely?
- Can we see errors early, not in production?
- Can we keep architecture clean over years?
However I admit AI is great with python so still ok for:
- quick scripts (through AI)
- experiments (POC / POCs Swarm through more advanced AI tools )
I am not sure I need to understand Python on a deep level. For my work, it is probably enough to know:
- basic syntax
- how to read simple Python code
- how to run a script and install packages
Most of the heavy work in Python can be done by AI tools. I just need enough Python to review the code, spot obvious problems, and connect it with the rest of my system.
Side note: Also, big blocks of modern monoliths can destroy your company in the AI world (or at least are a high risk), because you should aim to have components you can rewrite quickly - modules with clean interfaces. AI will struggle with monoliths and badly written legacy code. So small components are even more important than ever.