This project explores the potential of leveraging artificial intelligence (AI), specifically OpenAI's ChatGPT (version 3.5), to develop an API from scratch using only code generated by entering prompts. The goal was to investigate the extent to which AI could assist in the development process and evaluate the quality of the generated code.

The motivation behind this project stemmed from a curiosity about the capabilities of AI in software development. With advancements in natural language processing (NLP) and AI models like ChatGPT, it seemed plausible that AI could play a significant role in coding tasks, potentially speeding up development and reducing human effort.

The project began by experimenting with ChatGPT prompts to generate code for creating a basic API with user management functionalities such as user creation, login, and role assignment. Multiple iterations were conducted, refining the prompts, adjusting text inputs, and iterating on the generated code to achieve a usable API.

Several challenges were encountered during the development process:

1. Abstraction Level: As the project progressed and more complex features were added to the API, the AI struggled to maintain context and generate relevant code, leading to compatibility issues with existing code.
2. Deprecated Features: Due to the use of ChatGPT 3.5, the generated code relied on deprecated features, necessitating manual adjustments and potential issues with future compatibility.

Despite the challenges, the project yielded an API with basic user management functionalities. The generated code, while requiring refinement and manual adjustments, demonstrated the potential of AI in assisting with software development tasks. However, further improvements in AI models, such as transitioning to GPT-4, may address existing limitations and enhance code quality.

The project also involved comparing the performance of ChatGPT with Google's AI model, Gemini. Utilizing identical prompts for both models, it was observed that ChatGPT surpassed Gemini in generating usable code, highlighting the superiority of OpenAI's model in this context.

• Experimentation with newer AI models, such as GPT-4, to address limitations and improve code generation quality.
• Exploration of additional functionalities and complex features to assess the scalability and versatility of AI-driven development.
• Continuous monitoring of AI advancements and updates to leverage state-of-the-art technologies in software development projects.

The AI-driven API development project showcases the potential of leveraging AI, particularly ChatGPT, in assisting with software development tasks. While challenges exist, the results demonstrate promising capabilities and suggest opportunities for further exploration and refinement in AI-driven development processes.