AI tools in higher education: Ongoing misalignments

Abstract

Scholars have presented compelling arguments suggesting that a lack of artificial intelligence (AI) literacy (i.e. understanding the nature and capabilities of large language models (LLMs) like ChatGPT) is a primary driver of inappropriate use of LLMs (e.g. academic misconduct) by students. In contrast, I argue that more emphasis should be placed on the incentives for the misuse of LLMs. Specifically, the misalignments between the goals of key stakeholders (i.e. students/employers vs college faculty/administration) incentivize students to use LLMs in a manner that undermines the purpose of a college education, thereby demonstrating that AI literacy skills insufficiently prevent inappropriate LLM use. I describe the scope and implications of the misalignments prior to presenting pedagogical considerations for faculty and administrators. This presentation includes a recommendation that faculty and administrators utilize goal orientation to understand and mitigate the academic misconduct associated with incorporating LLMs into their respective pedagogies.