Code and Sensibility: ChatGPT in Humanities Research
Few developments in the last decade have impacted higher education so suddenly and dramatically as the December 2022 release of OpenAI’s chatbot ChatGPT. Responses from within higher-education ranged from eager adoption to outright hostility, accompanied by countless thought pieces attempting to predict the extent to which ChatGPT and its chatbot peers will upend our current model of higher education. While the debates about how higher-education will change are far from settled, there is little doubt that new AI technologies will force industries, whether in higher education or the public sector, to innovate at a rate faster than ever before.
One of the unique advantages of a school like Jessup — a school that both knows and equips its students — is that departments are small enough to innovate alongside the industries that are shaping our future. In the spring of 2023, the Jessup English department seized the opportunity to explore cutting-edge text analysis in a research practicum with several of our upper-division students.
Over the course of the semester, a group of five Jessup students and I worked together to learn the fundamentals of Natural Language Processing (NLP), a field that studies the intersection of computer language and human language, and discover how ChatGPT could aid us in writing code to perform textual analysis. The students, all English majors who were already masters of the art of close-reading, were now practicing interpreting texts at a thirty-thousand foot view, examining patterns that existed in dozens of texts across multiple decades.
As a group, we learned Python basics, used Jupyter Notebooks to execute and share our code, and discovered different Python libraries to help us perform our analyses and produce visualizations. ChatGPT was a massive aid in our journey, and even with our limited programming knowledge, we successfully used ChatGPT to iteratively compose the code we needed to perform our analyses. Our projects varied and produced an interesting range of insights: one project using clustering and classification algorithms on sets of texts; another learning how each algorithm would categorize texts from different eras; yet another examining patterns of co-occurring phrases across multiple texts. At the end of the term, the students presented their work to a small group of faculty and peers, outlining the journeys of their research projects.
While most of the students in the group are soon heading into graduate school, teaching, or editing, they now are equipped with some programming and data analysis skills — skills that are high on any list of most coveted skills for recent graduates. The future is bright for those graduates who can empathize, analyze, and innovate in equal measure, and it is in a liberal arts education that one can develop into an individual both broadly educated and eminently employable. The students I had the privilege of working with now could combine a humanist’s instincts with the tools of a data scientist. As George Anders explains in his wonderful book, You Can Do Anything: The Surprising Power of a “Useless” Liberal Arts Education, “The job market is quietly creating thousands of openings a week for people who can bring a humanist’s grace to our rapidly evolving high-tech future.”
Rather than rendering disciplines in the humanities obsolete, the recent strides in artificial intelligence have equipped the humanities with new tools to explore history and literature on a larger scale. Further, new technologies in AI have underscored how essential it is to know more than just programming and algorithms — in the future, some of the most interesting careers will belong to those who can adopt the humanist’s anthropological lens, attuned to different perspectives, rhetorical strategies, and the power of a good story.
And as it turns out, ChatGPT, oftentimes seen as a threat to disciplines like English and History, can be a wonderful tool to help those in the humanities do our work in exciting and innovative ways.
