Abstract:

The use of artificial intelligence (AI) tools is gaining momentum every day, becoming an indispensable resource for individuals across various age groups and professions. These tools serve both personal and increasingly business needs, offering quick feedback or responses to specific inquiries. Publicly available AI tools can be extremely accurate if used correctly. However, it is important to be aware of the limitations of AI tools and use them with caution. Thus, the use of AI tools has also found its use at all levels of education. AI can hardly completely replace the human factor in education, but it has a huge potential to transform education and make it more personalized, interactive and accessible to everyone.

This paper will compare test results of second-year undergraduate computer science students with those generated by two publicly accessible and free AI tools, ChatGPT and Gemini (until recently known as Bard). Specifically, the test results of students taking the "Introduction to databases" course will be compared to the results obtained using the aforementioned AI tools. The tests are exclusively quiz-type, offering multiple answers, with several correct answers possible. Also, in the case of choosing the wrong answers, the student receives negative points, and the minimum number of points students can achieve on the question is zero. The research employed identical tests for both students and AI tools.

The obtained results were analyzed to determine the difference between the students' results and the results obtained using the AI tools, and in order to ensure the validity of the results, the analysis includes statistical data processing and comparison of average points and percentages of correct answers. The results revealed significant discrepancies between student performance and AI-generated outcomes, with distinct patterns evident in the comparison. Overall, better results, as well as faster answers, were provided by AI tools. In some specific situations and for certain marginal cases, AI tools have shown shortcomings, especially when understanding the essence of the problem.

The research showed a comparison between the performance of students and AI tools in an educational context. While AI tools can augment traditional teaching methodologies, they cannot wholly replace human intervention. Further research is needed to find ways to integrate AI tools into the educational process that will maximize the benefits for students and teachers.

Keywords:

Artificial intelligence, ChatGPT, databases, education, Gemini, result comparison.