<![CDATA[The development of artificial intelligence technology has given birth to various applications that support the learning process, one of which is ChatGPT. The presence of ChatGPT makes it easier for students to find references, understand materials, and complete assignments. However, excessive use has the potential to create dependency, thus weakening students' learning independence. This study aims to analyze the level of student dependency on ChatGPT and its relationship to learning independence using a machine learning approach. The study was conducted by distributing questionnaires to 58 students covering seven categories of statements, including personal data, frequency of ChatGPT use, purpose of use, behavior towards ChatGPT answers, learning independence, attitudes towards dependency, and student dependency labels. The collected data were analyzed using the Naive Bayes algorithm with preprocessing processes such as data cleaning, coding categorical variables, and dividing the data into a training set (80%) and a testing set (20%). The results showed that the majority of students were in the medium dependency category (72%), followed by the low (13%) and high (15%) categories. The Naive Bayes model was able to classify the level of dependency with 83.3% accuracy on the test data. Further analysis revealed that the dominant factor influencing dependency was not only frequency of use, but also attitudinal aspects, such as a lack of confidence without ChatGPT and a tendency to use ChatGPT as a primary learning resource. It can be concluded that ChatGPT is an effective tool, but if not used wisely, it has the potential to undermine students' learning independence. Therefore, AI literacy and the cultivation of critical learning habits are needed so that this technology can support, not replace, the independent learning process.]]>
Copyrights © 2025
