Sunday, August 10, 2025

The Digital Press

All the Bits Fit to Print

Ruby
Web Development Artificial Intelligence
Urban Planning
Astronomy

LLM Tools Aid Coding but Struggle Beyond Popular Languages

An assessment of current large language model tools and workflows in software development.

From Hacker News Original Article Hacker News Discussion

A software developer spent about a month testing various AI coding tools and found that while large language models (LLMs) can be helpful, they have significant limitations, especially with less common languages and complex tasks.

Why it matters: LLMs can speed up certain coding tasks but often require strong existing codebases and developer oversight to be truly effective.

The big picture: Current LLMs favor popular languages and frameworks, struggle with complex or custom code, and demand constant updates to stay useful.

The stakes: Relying heavily on LLMs can degrade developer skills and create unstable workflows due to frequent model changes and unpredictable performance.

Commenters say: Many disagree with the article’s pessimism, emphasizing that effective LLM use requires adapting workflows and learning their quirks rather than outright dismissal.