Gemini AI Yielding Sloppy Code For Ubuntu Development With New Helper Script

Phoronix
Gemini AI produced sloppy, error-prone code when tasked with creating a helper script for Ubuntu's monthly ISO snapshot releases.

Summary

Following previous issues with Microsoft GitHub Copilot generating incorrect code for the Ubuntu Error Tracker, an Ubuntu developer tested Google's Gemini AI to create a helper script for the monthly Ubuntu snapshot releases, such as Ubuntu 26.04 "Resolute Raccoon" Snapshot 2. The developer found that Gemini produced similarly sloppy code, exhibiting issues like making silly mistakes, lacking semantic understanding, and creating confusingly named variables, suggesting that AI/LLMs still require significant improvement for large-scale software development practices. The resulting pull request shows the Gemini-generated code alongside the necessary subsequent revisions.

(Source:Phoronix)