Beware LLMs bearing gifts: hallucinations, satisfactory results, and student use of generative AI in the production of Mediterranean archaeology websites

Abstract

Abstract Recently, an epistemic shift has occurred in classics pedagogy. First, the COVID pandemic forced students online. Then, large language model (LLM) chatbots appeared. These disruptions created a conundrum: how could research methods, reading and writing, and cross-cultural exploration be taught when students were tethered to LLM-accessible ‘cheating machines’? The solution was AI-assisted classics. This paper describes how AI-assisted classical history can be taught through a discussion of the ‘WHAM’ (Web Design, HTML, and the Archaeology of the Mediterranean) project. WHAM is a website-building classroom assignment developed to teach students in the Computer Science and Society major at Wentworth Institute of Technology (Boston, MA, USA; www.wit.edu ) how to utilize digital tools to explicate humanities data. Students create a website from the HTML code up focused on a Mediterranean archaeological region and are taught to use LLMs critically, both for writing HTML code and for ideating web design features. In the process, students learn how LLMs must be approached cautiously since they can both hallucinate and offer satisfactory work that should be improved. Finally, the paper suggests that LLMs can provide students with a starting place for digital projects that challenge them to engage with the classical world’s history and material culture.