Benchmarking Generative AI Models for Classical Literary Criticism
By Joseph Dexter, Harvard University
Amid the daily torrent of news about the rise of AI, humanists in particular have stressed the need for critical evaluation of AI-generated content. To this end, classicists and ancient historians have already incorporated AI models into teaching to demonstrate not only the widely publicized occurrence of “hallucinations” but also infelicities of tone and expression (Bond 2023, Devereaux 2024). Rapid advances in large language models (LLMs), however, are opening up unprecedented possibilities for computational approaches even to complex and subjective tasks in classical studies.
AI, Machine Actionable Publication and Assigning Credit
By Gregory Crane, Tufts University
In 2023, Large Language Models (LLMs) developed the ability not only to translate but also to provide word-by-word linguistic analyses for historical languages such as Ancient Greek and Latin as well as Classical Persian and Middle Japanese. Users, including faculty and students alike, can now start reading and annotating sources in languages that they do not know, a process that involves interacting with, and comparing the results from, multiple LLMs.
Generative Image AI and Teaching Classics: A Case of Exaggeration
By Edward Ross, University of Reading
Generative artificial intelligence (AI) tools have become an integral part of daily life for teachers and students in Classics, whether or not everyone is aware. These tools, particularly generative image AI tools, have made it possible for teachers to produce images quickly that customize and diversify their teaching materials. Where these images would have previously been entirely restricted by the teachers’ skills and resources, the wide variety of increasingly effective generative image AI tools have made a far greater genre of images accessible, with some necessary caveats.
Prompt Engineering for Latin Teachers
By Patrick Burns, New York University
The sudden availability of chat interfaces for large-language models (LLM)—including ChatGPT, Gemini, and Claude, among others—has led teachers in all areas of education, including language instruction, to experiment with generating new or reworked materials for classroom use (Kohnke, Moorhouse, and Zou 2023). Latin language instruction is no exception (Ross 2023), as shown by workshops on LLM-based pedagogical applications that have been held on the topic (Burns 2024; Lamb 2024).
Opening Up Bottlenecks in Digital Classics Workflows with Human-in-the-Loop AI
By Samuel Huskey, University of Oklahoma
If you work on digital scholarship, the odds are good that you will encounter tedious and repetitive tasks. Depending on the scale of those tasks, you either resign yourself to grinding out the work manually, or you learn how to automate the tedious stuff, to paraphrase a primer on the subject (Sweigart 2020). Automating low-level tasks like cleaning and sorting data can accelerate workflows significantly, but higher-level tasks further down the pipeline require intelligent decision-making, which can lead to bottlenecks.