# The Humanities Can’t Ignore AI **By:** Robert B. Townsend **Published:** 2026-09-01T16:39:16.173Z **Source:** [The Chronicle of Higher Education](https://www.chronicle.com/article/the-humanities-cant-ignore-ai) --- Illustration by The Chronicle; Getty Like it or not, we need to learn the tools and their limitations. I get it. You hate AI. You think it destroys human capacities and the environment while also undermining critical aspects of humanities research and teaching. But if you care about your students and the future of the humanities, those objections are not sufficient. You need to engage with AI, understand both its appeal and its limitations, and integrate that awareness into your teaching. I get it. You hate AI. You think it destroys human capacities and the environment while also undermining critical aspects of humanities research and teaching. But if you care about your students and the future of the humanities, those objections are not sufficient. You need to engage with AI, understand both its appeal and its limitations, and integrate that awareness into your teaching. As the director of humanities, arts, and culture programs at the American Academy of Arts & Sciences — someone who works not only with academics but with artists, museum leaders, and technologists — I have been struck recently that faculty in the humanities seem the most hostile to AI. In meetings with artists and museum leaders, I find they often refer to AI as a “tool,” little different from a paintbrush or more recent software applications they have adopted into their work. The technologists, meanwhile, have their sights set on some distant horizon, casting magical-sounding solutions to big problems. So it is a bit jarring when I move from those conversations to discussions with faculty in the humanities, who often discuss the subject with an eyeroll and an assurance that they are now “back to blue books.” Recent [focus groups](https://www.amacad.org/publication/academic-humanities-today-opportunities-challenges-findings-conversations-department-chairs/section/8) with humanities-department chairs documented high levels of skepticism about AI, often in the face of pressure from administrators to swiftly take up those technologies. There are good reasons for suspicion about the integration of AI into the humanities, particularly given the ease with which output from a large language model can provide a superficial patina of expertise on any subject. If a student uncritically takes the output from an LLM as the extent of a research and writing task, that is obviously problematic. But we should also admit that many of the objections to AI overstate the comparative advantages of human work. I edited _Perspectives on History_, the American Historical Association’s magazine, for almost a decade and had to work through many of the same problems we now attribute to AI: basic factual errors, missing context, errant citations, confident assertions papering over logical and evidentiary gaps, and leaden prose. If we expected perfection from human beings, we would not have editors or peer review. So while convenient as a dodge, the problems with AI in its current form are not sufficient reasons for ignoring what it means for our disciplines. AI is intruding into every aspect of our work, whether we like it or not. If you want to articulate to your students to be on their guards and understand why they should be wary of those tools, you need to have some experience with the tools’ limitations. In the process, I think you will also find some good news, and evidence to show why a growing number of people in the AI industry and on the periphery of it [report](https://bit.ly/3St3rDE) they want to hire humanities majors to address many of the technology’s deficiencies. What are we training students to do if not identify questions worth asking, sharpen and frame those questions into research topics, gather information and critically assess its provenance and quality, and then synthesize all that information for a human audience? The skills students develop in humanities courses address many of the problems that quickly become evident with some informed engagement with an LLM. This is not a new problem. Over the past 40 years, I have watched humanities faculty incorporate new statistical tools, databases, word processors, digital sources, and other emerging technologies into their work. A few adopted the tools with enthusiasm, others warned about what it might mean for professional practices, and [most adopted them unevenly](https://www.historians.org/perspectives-article/historians-and-the-technologies-of-research-october-2017/) as particular uses made their work easier or opened up new research opportunities. Unfortunately, that uneven adoption leads to a partial and incomplete integration of these new tools into our work, as scholars often fail to fully grasp the potential strengths and limitations of each new tool for their research. Those problems then cascade into pedagogy, as students are often left to adopt those tools in a similarly haphazard fashion. So, what can humanities scholars do to familiarize themselves with AI? Start by asking ChatGPT and Claude to search the web and summarize your own work (with citations). Odds are that, like me, you will find they are not terrible at identifying patterns in your writing, but you are likely to spot a few hallucinations mixed in as well. The better you know a subject going in, the better able you are to spot problems and weaknesses in the LLM’s output and, more importantly, the spaces where training in the humanities can play a critical role in helping students be more discerning users of these tools. If you work with a couple LLMs in parallel, you will also begin to note important differences in their access to sources, chronological perspectives, and perception of what constitutes adequate research or prose. > There are good reasons for suspicion about the integration of AI into the humanities. But we should also admit that many of the objections to AI overstate the comparative advantages of human work. I recently used AI to dig through thousands of pages of notes and source material I had gathered more than 15 years ago, for a [book](https://press.uchicago.edu/ucp/books/book/chicago/H/bo14312740.html) on the history of the historical professions. While thinking about a new project on the ways historians adopted new technologies into their work in the past, I used it to scan my sources and it surfaced material I had completely forgotten about and accurately pointed out patterns — such as similarities in attitudes and language in sources written decades apart — that I had missed. Conversely, in that same scan of my source materials, I found the AI drew false inferences from historical documents because it failed to grasp irony and thus inverted the actual point of the author. In other cases, it failed to extract the most interesting material from documents because it was unable to “read against the grain” and capture critical hidden meanings. In each instance, I could recognize these problems only because I’ve spent years working with this material. All of this raises important questions about how to help students develop similar knowledge, experience, and disciplinary judgment, even when this new tool seems, to an untrained eye, to make those things unnecessary. Now is the time to think much more clearly about what we ask students to do and why. Simply dropping a prompt into an LLM and harvesting the results is not sufficient as either a beginning or an end to humanities work. The American Historical Association and Modern Language Association have started to articulate more detailed answers to these difficult questions, emphasizing AI literacy and the notion that “expertise must precede AI reliance.” But that leaves the hard work to each faculty member: translating their own specialized expertise into pedagogy and determining how to help students build up a foundational level of knowledge in their subject against this new background. Retreating to blue books and pretending AI does not exist is not the answer. So go ahead and despise LLMs all you want — but recognize that your ability to catch what they get wrong is the very skill set you want to cultivate in your students. It is on you to help them build it. _The views expressed here are solely those of the author._