Kira Rohwer
Last Name
Rohwer
First name
Kira
Email
kira.rohwer@unisg.ch
Phone
071 224 7547
3 results
Now showing 1 - 3 of 3
- Some of the metrics are blocked by yourconsent settings
Item type:Publication, ChatGPT’s Impact on Students’ Performance in Solving Tasks Requiring Critical ThinkingThis study investigates the impact of ChatGPT-assistance compared to web search with Google on undergraduate students’ task and learning performance in solving tasks that require critical thinking. A supervised online experiment with 39 undergraduate students from a Swiss business university was conducted. Participants completed 11 critical thinking tasks, deliberately designed to prevent ChatGPT from generating complete solutions in response to a single prompt, and a post-knowledge test to assess potential learning gains. The experimental group (EG, n = 20) used the generative AI tool ChatGPT-4.0 for assistance, while the control group (CG, n = 19) relied on conventional web search with Google. Welch’s t-tests revealed no significant differences between groups in rubric-rated task performance (p = .71) or post-test scores (p = .93). Nevertheless, performance in analytical tasks approached statistical significance (p = .06), while surprisingly moderately favoring the Google-group (d = 0.63). These results challenge the assumption that AI tools like ChatGPT inherently enhance task performance. Instead, they indicate that independent critical thinking remains essential for achieving performance gains in complex tasks. The findings highlight the need for higher education curricula and task designs that actively integrate AI tools while preserving students’ cognitive engagement, ensuring that they serve as a complement rather than a substitute for critical thinking.Type:conference paperJournal:Communications in Computer and Information Science (CCIS)Volume:2591 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Investigating the Division of Labour in Student-AI-Collaboration on Critical Thinking TasksThis study investigates the division of labor between students and generative AI in solving tasks addressing critical thinking. Three key research interests guide the study: (1) the extent of ChatGPT usage by students, (2) the division of labour between student and AI as perceived by students, ChatGPT, and an expert, and (3) if these ratings align with each other. An experiment involving 20 undergraduate students who completed 11 tasks using ChatGPT was carried out. Findings demonstrated substantial variation in the extent of ChatGPT usage among students. The division of labour analysis revealed that ChatGPT is used most heavily for reasoning tasks, and that students consistently rated the AI's contribution to their task solutions as slightly lower than both ChatGPT and the expert. Correlations of the ratings indicated strong alignment between students and ChatGPT, but weaker alignment with the expert perspective, particularly in the metacognition domain. These results highlight the need for enhanced integration of AI in educational practices, focusing on fostering students' understanding of AI capabilities while promoting critical engagement. Future research should explore diverse populations, a broader range of task types, and the long-term impact of AI integration to enhance the effective use of generative AI in education.Type:conference paperJournal:2025 IEEE International Conference on Advanced Learning Technologies (ICALT) - Some of the metrics are blocked by yourconsent settings
Item type:Publication, DEVELOPING A FRAMEWORK FOR ANALYSING AND ASSESSING CRITICAL THINKING SKILLS FOR THE RESPONSIBLE USE OF GENERATIVE AI IN HIGHER EDUCATIONCritical thinking skills are essential in the age of Artificial Intelligence (AI) to ensure that students can effectively evaluate AI-generated content, avoid bias, and make informed decisions that meet ethical standards in higher education. Based on the foundational literature on critical thinking [1], the study proposes a framework for analysing and assessing critical thinking skills with a focus on the appropriate use of generative AI. The framework consists of four dimensions: 1) analytical skills: the ability to interpret and understand the output of ChatGPT, recognising biases and underlying assumptions, 2) evaluation skills: evaluating AI-generated text, data, images, 3) ethical reasoning skills: reasoning and decision making, and 4) metacognition: reflecting on personal biases, reflecting on ChatGPT use. On this basis, we developed a task-set compromising of 11 critical thinking tasks and conducted an online experiment with 39 students. The experimental group (EG, 20 students) used ChatGPT for assistance, while the control group (CG, 19 students) relied on conventional web search in Google. Subsequently, students were asked to evaluate the tasks in terms of 1) ease of understanding, 2) ease of solving, and 3) satisfaction with their own performance. The results demonstrate that the task scores of the EG are, on average, slightly higher than those of the CG. However, the results show shortcomings of the EG in the field of analytical skills, deductive reasoning and the reflection on the appropriate use of ChatGPT. The study is limited by the small sample size and further research is needed.Type:conference paperJournal:ICERI2024 Proceedings