Hauptinhalt

Pros and cons of artificial intelligence as a tool for constructing and assessing biology assessment (Prof. Dr. Zbigniew Adamski)

Veranstaltungsdaten

04. Februar 2027 09:00 – 04. Februar 2027 11:00
Termin herunterladen (.ics)

BBB

Join online via BBB: https://webconf.hrz.uni-marburg.de/n/rooms/1ey-vvp-fjj-szb/join 

Rapidly developing artificial intelligence (AI) is gaining increasing interest in education. It can be used as a tool for self-study, personalized learning, creating and grading tests. This makes it a tempting tool for students and teachers alike, as it can reduce the time spent grading or completing assignments. However, research suggests that AI can make errors. We decided to test how a simple, free version of AI would perform as a tool for: 1) solving high school biology tasks; 2) whether the AI ​​would be able to correctly assess students' answers; 3) whether the AI ​​would correctly construct the criteria for answers used to evaluate student responses.

Our research relied on materials from the official Matura exams in biology. In Poland, this exam is standardized and uniform across the country. It is assessed by examiners: high school and university teachers. Exam results are also treated as grades that qualify students for university admission.

In our study, the answers and grades provided by AI were compared with the grades assigned by three experienced examiners. The results indicate that AI models can be useful in grading student papers and for independent study, but they should be treated with caution. AI models can solve tests at a good, but not perfect, level. Student assessment of answers is also not perfect. Similarly, when it comes to grading criteria, AI creates them less precisely than the team that designed the Matura exam. Furthermore, the level of accuracy of answers depends largely on the question type.

In this presentation, we would like to present our observations on the possibilities and threats of using artificial intelligence in teaching at the secondary school level, the possibilities of using artificial intelligence to create and assess tasks, and, consequently, as a tool for self-learning.

Referierende

Prof. Dr. Zbigniew Adamski, Adam Mickiewicz University, Poznań, Poland

Kontakt