ChatGPT is not a pocket calculator -- Problems of AI-chatbots for teaching Geography
Project Overview
The document explores the implications of generative AI, particularly ChatGPT, in the realm of education, focusing on its application in teaching and assessing Geography and GIScience. It highlights the potential benefits of generative AI, including personalized learning experiences and a reduction in teachers' workload. However, it also raises significant concerns regarding the validity of assessments, as AI's capacity to generate high-quality responses can obscure the accurate measurement of students' skills and knowledge, particularly for tasks requiring higher-order thinking as outlined in Bloom's taxonomy. The authors emphasize the need for a reevaluation of assessment strategies to accommodate the capabilities of AI, advocating for adaptations to traditional assessment methods to preserve educational integrity and ensure that evaluations truly reflect student learning outcomes amidst the challenges posed by advanced AI technologies.
Key Applications
ChatGPT for generating answers to Geography and GIScience exam questions
Context: University-level Geography and GIScience courses
Implementation: Survey conducted among university teachers who posed exam questions to ChatGPT and rated the responses
Outcomes: ChatGPT responses were rated as correct enough to pass exams (80% correctness), but issues with completeness and conciseness were noted.
Challenges: Concerns about the validity of assessments due to AI's ability to generate plausible responses that might mislead evaluators.
Implementation Barriers
Ethical
Use of AI chatbots can lead to academic dishonesty by enabling students to submit work that does not accurately represent their knowledge and skills.
Proposed Solutions: Implement strict assessment controls, adapt learning goals, closely supervise learning processes, or forbid the use of chatbots in assessments.
Project Team
Simon Scheider
Researcher
Harm Bartholomeus
Researcher
Judith Verstegen
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Simon Scheider, Harm Bartholomeus, Judith Verstegen
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai