Prompt Engineering in User Interface DevelopmentDelić, Faris (Avtor) Lavbič, Dejan (Mentor) Artificial intelligencePrompt engineeringUser interfaceLarge language modelsCode qualitySoftware developmentThis diploma thesis addresses the growing demand for efficient software development processes in the era of Artificial Intelligence (AI). A systematic approach called Prompt Engineering involves designing and refining prompts to optimize AI-driven code generation. My approach leverages AI models to generate user interface (UI) code for web applications, significantly reducing development time and improving code quality. The thesis evaluated and compared leading LLMs in today’s world (ChatGPT, Gemini, and Phind), while providing them with prompts of varying quality. The models were tasked with creating a website for displaying information about countries of the world. The best performance was shown by ChatGPT, which excelled at every level of prompt quality. The takeaway is that Phind and Gemini require more prior knowledge to be used effectively, while ChatGPT generates better results even when users are not completely certain about what they need. Furthermore, citing sources by Phind and Gemini is commendable but is in fact the only advantage these two models have over ChatGPT. The results indicate that the use of AI, specifically LLMs, can significantly improve performance and efficiency in user interface development, especially if the user has prior knowledge.20242024-09-06 16:25:06Diplomsko delo/naloga161083VisID: 37306COBISS_ID: 211168003sl