From Natural Language to Broken Layout: A Pro...

From Natural Language to Broken Layout: A Prompt Driven Failure Analysis of AI-Generated Frontend Code a Comprehensive Research Study Comparing ChatGPT, Gemini, and Claude

Author Name : Mayank Dalal, Dr. Rahul

ABSTRACT This paper investigates the limitations and different types of errors in the code written by LLMs after receiving natural language prompts. The widespread use of such models like ChatGPT, Gemini, and Claude in the software development sphere has made frontend code generation one of the main applications of these models. Despite compliance with syntactic standards, the output results of LLMs could be characterized by poor visual consistency, semantic and functional mistakes.