Abstract
This study explores the potential of chatbots, specifically ChatGPT, in Java software development. The aim is to classify tasks for effective use of industrial code and develop recommendations for applying chatbot assistance, identifying boundaries where human intervention remains essential. The methodology included analyzing scientific literature and empirically testing ChatGPT-3.5 on various Java development tasks. The tasks were divided into simple (working with XML, JSON, multithreading, and data input/output) and complex (writing MVC applications, REST services, and GUI). The results showed that ChatGPT successfully handles simple tasks but struggles with complex problems. The study identified scenarios where the chatbot can effectively use existing codebases and design patterns to accelerate development. The conclusions highlight ChatGPT’s potential in improving developer productivity, optimizing certain development tasks, and more efficiently allocating human resources in projects. However, the study also points out the need for human intervention to verify, correct, and improve generated code. The study contributes to understanding the practical usefulness of chatbots in real development scenarios and offers recommendations for integrating AI tools into the software development process.