JAVA WEB PROGRAMMING
WITH CHATGPT
N A M E : AV I N E S H
REG NO:22MIS0522
JOURNAL PAPER
Year Authors Paper Title Publication Details Link
International Journal https://ieeexplore-
Java Web of Computer ieee-
2024 Muhan Guo Programming with Applications (IJCA), org.egateway.vit.ac.i
ChatGPT Volume 177, Issue n/document/
39, April 2024 10629560
ABSTRACT
With the growth of AI, ChatGPT has become a major breakthrough in
natural language processing. It is widely used for conversation generation
and now brings both opportunities and challenges to Java Web
programming. Unlike earlier studies focused on small coding tasks, this
paper tests ChatGPT’s ability to generate complete Java Web projects
using a user-login case. The results show that the code is readable,
functional, and of good quality, but performance drops when instructions
are unclear. Overall, ChatGPT can be a helpful tool for developers,
improving efficiency and productivity
L I T E R AT U R E S U RV E Y
Authors Paper Title Description Year DOI / Link
Built a student
management system
A Study of Web Code
using ChatGPT-4o; tested
Zhan Shu, Zijie Dong Generation Based on 2025 DOI: 10.4108/eai.21-11-20
functionality, accuracy, 24.2354633
ChatGPT
human involvement, and
limitations in web dev.
Examines ChatGPT-
A Pilot Study on Secure generated web app code;
Code Generation with studies vulnerabilities
(ACM Authors) 2024 ACM Link
ChatGPT for Web (SQLi, XSS) and ways to
Applications refine prompts to improve
security.
Compares AI vs human
solutions in
Baskhad Idrisov, Tim Program Code Generation Java/Python/C++;
2024 DOI: 10.3390/a17020062
Schlippe with Generative AIs evaluates correctness,
efficiency, and
maintainability.
Compares GPT-3.5, GPT-4,
Rushabh Jain, Jyoti The evolution of ChatGPT GPT-4o on algorithmic
Thanvi, Akila for programming: a problems (Java, Python, 2025 DOI: 10.1088/2631-8695/a
Subasinghe comparative study C++); analyzes runtime, da51d
memory, complexity.
Mafura Uandykova,
Laura Baitenova, Discusses AI’s role in Java
Gulnar Java coding using artificial coding, trends, challenges,
2024 DOI: 10.3389/fcomp.2024.
Mukhamejanova, Assel intelligence and applications in 1473870
Yeleukulova, Tolkyn programming tasks. 4
Mirkassimova
ARCHITECTURE DIAGRAM
METHODS / PROPOSED SYSTEM
•Objective: Test ChatGPT’s ability to generate Java Web code (client–server).
•Case Study: User Login System → login.html → VerifyServlet.java → welcome.html.
•Process:
1.Divide project into small tasks.
2.Prompt ChatGPT for each step (HTML, Servlet, Result page).
3.Deploy in Tomcat & test via Chrome.
•Tools: ChatGPT-3.5, IntelliJ IDEA, Tomcat 8.5, Servlet-api.jar.
•Evaluation: Checked code readability (CheckStyle), performance & memory
(VisualVM).
•Result: Code was readable, functional, efficient, but struggled with vague
instructions.
S O F T WA R E / D ATA S E T U S E D
Software:
•ChatGPT-3.5: The AI language model used to generate the code.
•IntelliJ IDEA-2020.1.2: The integrated development environment (IDE) for building and testing the web project.
•Tomcat-8.5: The web server where the Java application was deployed.
•Google Chrome: The web browser for user interaction.
•Servlet-api.jar: A library file essential for handling web requests and responses.
•CheckStyle: An automated tool for verifying code quality and conventions.
•VisualVM: A visualization tool for monitoring the program's performance, including CPU and memory usage.
Dataset:
•User-login study case: This was the primary case study, consisting of three components: login.html, a VerifyServlet.java
file, and welcome.html. The purpose was to create a functional login system to test the interaction between the client-side
and server-side code.
7
R E S U LT
I. Results Overview
The project was fully functional, successfully handling the interaction between the client-
side pages
The generated code was characterized by high readability, good quality, and complete
functionality.
The study found that ChatGPT's performance was negatively affected when given
instructions with limited text information.
II. Key Performance Metrics
Code Readability: An automatic code review tool, CheckStyle, confirmed that the
generated code followed standard coding conventions, with no errors reported. The code
also included explicit comments to improve readability and maintainability.
Memory Utilization: The program used a minimal amount of memory, consuming only
8,208 bytes. Monitoring with VisualVM showed low heap memory usage (15-22 MB) and no
memory leaks.
Runtime Performance VerifyServlet executed in just 153,900 nanoseconds,
demonstrating a very quick response speed.
8
9
CONCLUSION
This paper evaluates how well ChatGPT can generate code for Java web problems, using a user-login program
as a case study. The experiments show that the project built from the generated code meets all expected
requirements. The code is noted for its
high readability, good quality, and complete functionality. The study concludes that ChatGPT can be a rapid
and effective assistant for developers, offering a solution that is faster than manual coding.
Future work will focus on exploring ChatGPT's ability to generate code for more complex, large-scale Java
enterprise applications and other programming languages
FUTURE WORK
Exploring Code Generation for Complex Applications: Investigating ChatGPT's ability to generate code for more complex,
large-scale Java enterprise applications.
Utilizing More Quality Metrics: Employing a wider range of metrics to evaluate the quality of the generated code more
thoroughly.
Solving Problems in Other Programming Languages: Extending the research to explore ChatGPT's capabilities in solving
problems in programming languages other than Java.
THANK YOU
