Përdorimi i intelegjencës artificiale për gjenerimin automatik të kodit në front-end development

Abstract

The use of artificial intelligence (AI), especially large language models and multimodal architectures, has brought a profound revolution in the way user interfaces are built and developed in front-end development. This study focuses on the use of AI for automatic code generation in the development of the front-end part of web applications, with special emphasis on the process of converting visual design into functional code as well as intelligent assistance through autocomplete systems and contextual suggestions in modern IDEs. Advanced experiments, such as those carried out in the Design2Code study, have shown that models like GPT-4V are capable of recreating existing web pages with high visual and structural accuracy, achieving up to 49% accuracy in reproducing the original appearance and over 60% in user-perceived correctness. However, these systems still face challenges related to element hierarchy, pixel-perfect positioning, and semantic coherence of HTML, CSS, and JavaScript code. In this context, more sophisticated approaches have been proposed, such as WAFFLE, UICopilot, and DesignCoder, which use customized training, self-correcting mechanisms, and hierarchical structures to significantly improve the quality of generated code. On the other hand, the integration of AI-powered assistant tools into development environments, such as GitHub Copilot and intelligent IDEs like Cursor, has transformed the way developers interact with code, offering not only autocomplete but also personalized recommendations, code refactoring, and even intelligent debugging. The adoption of these technologies has increased significantly in the software industry, shifting from partial usage to full assistance during development and testing phases, and in some cases even automating deployment. However, this approach is not without risks: recent studies have revealed that a significant percentage of AI-generated code contains known security vulnerabilities, such as XSS and SQL injection, while developers themselves have reported often feeling uncertain about the correctness and accuracy of the output provided by these systems. This highlights the need for rigorous code quality checks, the integration of automated security assessments, and the building of greater awareness of safe coding practices when using AI.The use of AI for automatic code generation in front-end development represents a major step toward the full automation of the software development cycle, but for successful and sustainable adoption, a balanced approach between technological innovation and human careful control over quality, security, and development responsibility is essential.