# LLMs' Reshaping of People, Processes, Products, and Society in Software Development: A Comprehensive Exploration with Early Adopters

**Authors:** Benyamin Tabarsi, Heidi Reichert, Ally Limke, Sandeep Kuttal, Tiffany Barnes  
**arXiv:** [2503.05012](https://arxiv.org/abs/2503.05012)  
**Date:** 2025-03-06  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2503.05012v1/

## Key points

- Large language models (LLMs) like OpenAI ChatGPT, Google Gemini, and GitHub Copilot are gaining traction in software development.
- Lack of in-depth qualitative assessments on how LLMs are utilized in real-world software engineering contexts.
- Study conducted with sixteen early-adopter professional developers to explore the impact of LLMs across various stages of the software development life cycle.
- Four key dimensions examined: people, process, product, and society.
- LLMs greatly enhance routine coding tasks such as code generation, refactoring, and debugging.
- Clear problem statements lead to optimal results when using LLMs.
- Significant values for using LLMs include facilitating learning new languages and concepts within software engineering.

## Summary

In the rapidly evolving landscape of software development, large language models (LLMs) such as OpenAI ChatGPT, Google Gemini, and GitHub Copilot are gaining significant traction. Despite their increasing adoption, there is a lack of in-depth qualitative assessments on how LLMs are utilized in real-world software engineering contexts. To address this gap, a study was conducted involving semi-structured interviews with sixteen early-adopter professional developers. The research focused on exploring the impact of LLMs across various stages of the software development life cycle, examining four key dimensions: people, process, product, and society. The findings from the investigation revealed that while LLMs have not brought about a fundamental revolution in the development process, they have greatly enhanced routine coding tasks such as code generation, refactoring, and debugging. Developers reported optimal results when providing clear problem statements to LLMs, highlighting their effectiveness with decomposed problems and specific requirements. Additionally, were identified as significant values for using LLMs among early adopters. They facilitate learning new languages and concepts within the field of software engineering. persist despite the observed benefits by experts in both software engineering and LLM functionality.