We are living through a period described by technologists as an “AI Boom” or “AI Spring.” A swift and impressive gain of function in generative artificial-intelligence systems (AI), made possible by developments in computer-science techniques known as deep neural networking, has given new force to old questions about technology and society. Drawing on new designs for large language models especially, technologists have sustained a period of major investment and growth in the industry.
The advent of specific AI tools, especially ChatGPT and image generators, has led professors and students of the humanities, as well as tech-resistant artists and writers, to wonder aloud what awaits the humanities in the age of AI. Much of the debate around this question focuses on the state of teaching, grading, and writing. These themes encourage humanists to interpret AI using the realpolitik of professional jurisdiction, the weighty terms of academic ethics, and the dull drama of arcane university policy.
Some scholars and writers reject AI technology outright as antithetical to their pursuits as humanists.Some scholars and writers reject the new technology outright as antithetical to their pursuits as humanists, assuming that AI represents vulgar techniques that cheapen the craft or noisy machines that disturb the garden. Amidst the handwringing, however, a cutting-edge team of researchers has harnessed AI to generate field-shaping discoveries promising to shore up humanistic pursuits within the Ivory Tower.
When ChatGPT arrived on campuses in 2022-23, administrators and faculty openly debated their genuine concerns about a “crisis” of education in the age of AI. Was student use of the chatbot cheating or simply another online tool? Perhaps bearing in mind perceptions concerning the suspect value of a humanities degree, not to mention general trends of stalled or actively declining numbers of undergraduate humanities majors, some scholars warned that, absent precautions, the horizon of humanistic study would soon be clouded by darkness. In the New Yorker, Nathan Heller heralded the “end of the English major.” As students instructed ChatGPT to complete coursework, and as ChatGPT composed papers approximating the prose of a competent—and perhaps even an insightful—student, few could blame Heller for sounding the alarm.
Meanwhile, some educators and professors took to ChatGPT to draft syllabi, craft lesson plans, and grade student papers. As many institutions of higher education adopted bans on ChatGPT in the classroom in early 2023, one Reddit thread posed the question, “Why are teachers being allowed to use AI to grade papers, without actually reading it, but students get in trouble for generating it, without actually writing it?” For a moment, it appeared that higher education might become, for some, a feedback loop of machine-machine communication.
Loftier questions about AI’s meaning for humanity loomed large: questions about safety, existential risk, automation, and harmful bias. Institutions of higher education were theaters to enact answers to these questions, and scientists, humanists, and students alike played actors upon the stage. Universities themselves crafted and then repealed polices about ChatGPT in the classroom. Some scholars and humanists saw AI as anti-human, while others saw it as merely another tool for humanistic work.
The specter of AI running amok, turning us all into paperclips, and the challenge of chatbots in the classroom sucked up most of the oxygen in debates about AI in higher education. On the margins of academia, however, some researchers have sought to use new AI technology to unlock secrets about the past, even secrets locked in carbonized Roman scrolls that cannot be handled by humans without significant risk of destruction.
On the margins of academia, some researchers have sought to use AI technology to unlock secrets about the past.Of the artifacts from antiquity that still puzzle academics, the Herculaneum scrolls rank among the most difficult. The scrolls date to A.D. 79, when Mount Vesuvius violently erupted, showering settlements like Pompeii with pumice stones, volcanic ash, and flows of pyroclastic sludge that encased the bodies of the fallen. In nearby Herculaneum, a villa containing about 1,000 papyrus scrolls and scores of relics was entombed in the same cataclysmic eruption. The papyri may have been owned by Julius Caesar’s father-in-law, but scholars are unsure. Researchers are certain, however, that the scrolls are collectively one of the oldest libraries known to man. When a farmer discovered them in the 1700s, the scrolls became vexing artifacts of academic endeavor—objects brittle to the touch and scorning, in their very form, attempts to decipher their contents. Many fell apart, breaking into ashes. 270 scrolls remain today, held in Naples, containing text that has gone unread for millennia.
Until recently, that is. In March 2023, University of Kentucky computer scientist Brent Seales and entrepreneurs Nat Friedman and Daniel Gross launched the Vesuvius Challenge. The challenge equipped competitors with new software and thousands of 3D, X-ray images of the Herculaneum scrolls and offered $1 million in prizes to researchers. The grand prize, a whopping $700,000, would go to competitors who deciphered, at a minimum of 85-percent readability, four passages from scroll images, with each passage totaling a minimum of 140 characters.
By February of 2024, the challenge announced a winning team of three: Luke Farritor, a computer-science student in Nebraska; Youssef Nader, an Egyptian Ph.D. student in Germany; and Julian Schilliger, a robotics student in Switzerland. (Farritor also won the challenge’s “first letters” prize for deciphering purple, or “porphyras” in ancient Greek, as the first legible word on the scroll used for the competition.) Together, the team used machine-learning algorithms to crack the translations of more than 2,000 characters, far surpassing the minimum. In total, the team’s discovery represents only five percent of the scroll’s text.
Challenge judge and University of Naples papyrologist Federica Nicolardi noted that the judges “were all completely amazed” by the “incredible” results. On NBC News, the distinguished University of Michigan classicist Richard Janko described the challenge as a watershed moment. “Until now, the only way we have had to read what’s inside the Herculaneum scrolls is to put together the thousands of pieces of the ones that crumbled apart,” Janko said. “It’s like putting together a mosaic, and there’s not many people willing to do it.” Therefore, he added, “it may take 500 years to decipher their content.” Given the machine-learning technique developed through the challenge, however, “hopefully, it should be much easier, and quicker.”
The humanities in the age of AI will not be relegated to a footnote.Later this year, teams will compete in Vesuvius Challenge Stage 2. In this competition, a grand prize will go to the first team to decipher 90 percent of not just one but four scrolls. If teams succeed, University of Bristol classicist Robert Fowler suggests, the discoverers may “rewrite the history of key periods of the ancient world.”
Humanists do not need to share Silicon Valley’s rampant techno-optimism. The nearly religious faith expressed by leading technologists like OpenAI’s on-again, off-again CEO Sam Altman, who handwaves at concerns about AI and states that the future is “going to be great,” is merely sleek copy for venture capitalists.
But the humanities in the age of AI will not be relegated to a footnote. We should recognize that AI is not merely an army of chatbots devaluing writing, the coin of humanistic inquiry. Rather, AI is a tool, and some craftsmen are already at work, using it to uncover new texts for humanists to interpret, debate, and share with the world.
Jacob Bruggeman is a Ph.D. candidate in history at Johns Hopkins University, where he is the inaugural graduate fellow of the Center for Economy and Society and an associate fellow at the Institute for Applied Economics, Global Health, and the Study of Business Enterprise. He is associate editor of FUSION and editor-at-large at the Cleveland Review of Books. He is a graduate of Miami University and the University of Cambridge, Darwin College.