Akademik Veri Yönetim Sistemi
DINBILIMLERI AKADEMIK ARASTIRMA DERGISI-JOURNAL OF ACADEMIC RESEARCH IN RELIGIOUS SCIENCES, cilt.24, sa.3, ss.29-54, 2024 (ESCI, TRDizin)
This study aims to analyze the reliability and accuracy of responses provided by artificial intelligence-based conversational agents, specifically ChatGPT and Google Gemini, to exam questions in Islamic Property Law. Artificial intelligence, as a technology mimicking human intelligence and making significant advances in language processing, holds vast potential across various domains, including software, medical diagnosis and data analysis, digital assistance, ecommerce, gaming, translation, education, data forecasting, recommendation systems, cybersecurity, and information management. Recently, AI-based conversational agents have gained substantial attention due to their capacity to respond to complex information requests, an area advanced through research on language models. Within this context, advanced and popular language models such as ChatGPT and Google Gemini are considered valuable information sources for educational and exam preparation processes, facilitating and expediting student access to information. Particularly, ChatGPT, developed by OpenAI and trained on extensive datasets, is recognized for its high-accuracy responses across diverse fields due to its natural language processing capabilities, resembling human intelligence, and its capacity to interpret contextual meanings. This ability makes ChatGPT a quickly accessible resource for students in fields such as education, information access, and guidance. Meanwhile, Gemini, developed by Google AI, stands out for its real-time internet access, providing users with updated information and continuously refreshing its knowledge base, positioning it as a competitor to ChatGPT with its ever-expanding pool of accessible data. The study focuses on Islamic Property Law due to its complex and comprehensive nature within Islamic law, as it meticulously regulates individuals property rights. This field addresses processes such as the acquisition, utilization, transfer, and termination of ownership, aiming to protect individuals' rights over their assets in accordance with Islamic legal principles. Islamic Property Law encompasses fundamental concepts like real and personal rights, definitions of ownership, types of ownership, and the authority individuals have over their assets. By not only safeguarding individual rights but also defining the limits of ownership and conditions for valid transactions, it functions as a mechanism to prevent the misuse of these rights. Another significant aspect of Islamic Property Law is its aim to ensure social justice by developing concepts such as the rights of Allah, going beyond the protection of individual property rights. Emphasizing that property should serve the public interest, this area of law envisions ownership within a framework that considers public welfare. This approach supports societal stability while ensuring balanced protection of individual rights. Thus, Islamic Property Law aims to promote the welfare of not only individuals but also society as a whole, balancing personal property rights with socially beneficial use. Due to its intricate and sophisticated structure compared to other domains, Islamic Property Law was specifically chosen for this study to comprehensively evaluate the response capabilities of artificial intelligence. Its unique and detailed nature provides a suitable foundation for assessing AI models' capacity to respond to complex legal concepts. Accordingly, this study seeks to evaluate the reliability of responses provided by ChatGPT and Google Gemini to Islamic Property Law questions and to assess their potential to enhance students' exam performance. In this evaluation, various exam questions covering fundamental concepts of Islamic Property Law and classified into easy, medium, and hard levels according to students' knowledge were posed to both AI models. These questions pertain to topics such as property rights, asset status, and conditions of use, aiming to measure students' depth of knowledge in Islamic Property Law. To more comprehensively analyze the reliability of the responses, statistical examinations were conducted using SPSS software. The normality of score distributions was assessed using the Shapiro-Wilk test, enabling the application of parametric tests. Additionally, an independent sample t-test was applied to determine scoring consistency among researchers, with results indicating aligned scoring trends without significant differences. To further strengthen the consistency of the evaluation process, Spearman's rho correlation coefficient was calculated, revealing a high degree of agreement among researchers. Lastly, the word count and average scores of the responses were analyzed via SPSS, establishing a solid methodological framework for a comparative assessment of the reliability and accuracy of responses provided by ChatGPT and Google Gemini to Islamic Property Law exam questions. The study's primary findings shed light on the capacity of AI-based conversational agents like ChatGPT and Google Gemini to deliver information in specialized fields such as Islamic Property Law. Although both models offer advantages in providing rapid access to information during exam preparation, the accuracy and comprehensiveness of their responses in areas requiring extensive interpretation and expertise, such as Islamic law, can be limited. As the accuracy of AI-provided information directly affects student exam performance in complex fields such as law, careful evaluations of the reliability of AI chatbots in these areas are essential. This study's reliance on evaluations from only two Islamic law experts may introduce a limited perspective; however, this limitation also allows for broader interpretation of the results. Future studies involving a more extensive expert panel, along with diverse academic approaches and testing models, will provide a more comprehensive examination of the reliability and efficacy of AI tools in the domain of Islamic Property Law. Such research can offer valuable insights into determining the proper uses of AI in this field. However, the extensive use of AI models risks distancing students from primary reference sources, particularly core textbooks. This may lead to superficial presentation of information in digital environments, detached from its context, potentially resulting in incomplete and unreliable knowledge. Therefore, while AI facilitates rapid and easy access to legal information, the accuracy of the responses provided must be carefully evaluated by Islamic law experts..