• Challenging AI as Critical Thinking

    Michael S. Kirkpatrick

    Abstract

    Generative AI systems, such as LLMs, have rapidly become popular tools for students and professionals alike. While the results of these systems can often be useful due to their underlying statistical properties, these tools can produce erroneous and harmful results if not evaluated critically. In this work, we discuss the nature of LLMs, their potential benefits for education, and their flaws. In particular, we highlight both epistemological and moral concerns that users should heed when using them. To address these concerns, instructors should model how to interact with these tools as a form of critical thinking. Our goal in this article is to provide the necessary context and potential directions for using these tools successfully to augment the social process of education.