Abstract

A panel on which skills graduate programs do not teach and what departments can do now, including methods training and hiring. Jacob Bruggeman argues that debate over AI in general misleads and that graduate students should judge specific agentic tools, meaning systems that carry out multi-step tasks such as optical character recognition (OCR) on their own, case by case. Amanda Regan describes the Clemson University PhD in digital history, where students learn to code by hand and must explain their code, and Christopher Phillips argues that every historian already does computational history. Sarah Weicksel describes the American Council of Learned Societies (ACLS) Doctoral Futures project, and the audience asks whether sole authorship sets the bar too high.

Chapters

Transcript

AI-generated transcript. It was produced from the recording by speech recognition and cleaned up with AI. It has not been edited by the speaker and may contain errors. The captions on YouTube are AI-generated in the same way. Each timestamp links to that moment in the video.

[0:05] Post-lunch conversations. Grab some coffee, if you haven't already. And we'll talk about grad education in the age of computational history or computational history in the age of agentic AI. Okay, so my name is Jacob Bruggeman. I'm an assistant professor at Ohio State. I'm a historian of cybersecurity and computing. I finished my PhD here in July 2026 and just moved over. So take everything I say as a professor. from somebody with no experience teaching or training graduate students, but somebody who came of age before, I guess, entered a PhD program before agentic AI, LLMs really struck the academy and then sort of finished and entered a job market, both skilled up in this area and also in a market kind of defined by worries about it. Okay. So like many of us in 2023, I was confronted with, you know, ChatGPT appearing in the classroom, worrying about what

[1:05] perhaps the future for historians like me would look like a historian of computing, but not necessarily a born and raised programmer. Right. And these fears were only accentuated when, in 2025, Microsoft released a report saying that top 40 occupations, the highest AI applicability score. And lo and behold, historians, writers and authors are number two and three, respectively, in this report. So my initial impression, of course, is I'm going on the market. What the heck is this all about? I hate these tools like the rest of America, when I think combined with worries about tech fascism, so-called tech fascism, the authoritarian turn with the way that tech leaders seem so dismissive of humanistic enterprise. I think it's easy to want to resist and push back to AI tools. But as I experimented with them, drew from my own experience and background expertise in the history of computing, I think that this initial impulse to seeing

[2:08] Microsoft and other firms so readily claim their ability to automate what we do, this initial impulse faded into more curiosity about how these tools can change, accelerate, maybe even make better the things we already do. So the current debate in the profession, as I understand it, focuses a lot on, writ large, on the political elements of AI and AI tools, right? The intellectual potential errors or problems that emerge from relying on them, right? The hallucinations that might slip into our manuscripts, the cognitive surrender we might be performing if we integrate them without too much care. Or the slop, right? Using these tools will inevitably lead you to producing historical slop. I think this debate about general AI in history has been misleading. And what we need to really be focusing on when it comes to graduate training, when it comes to people entering the profession, is specific arguments about how AI tools relate to our work as historians.

[3:09] So I'm going to talk about three of them very briefly. I think all incoming graduate students and junior folks really need to be thinking about AI tools as a sort of warm knife to cut through the butter of bureaucracy at their institutions, but not in a DOGE, crazy DOGE way. How they can accelerate and implement better research practices for us. And I think also reignite some of the things that got us interested in doing history in the first place. Intellectual ambitions. There's also a downside to this too, we'll conclude with. So bottom line, I think a lot of the discussion in the Chronicle in places like it is all or nothing. It's driven by political, intellectual, aesthetic commitments that are unlikely to be changed in a conference like this. But if we talk about tools at a case-by-case basis, I think we can very clearly see the trade-offs and the advantages for working with rather than against AI. Okay, so my own view by AI here, I'm going to talk specifically about agentic systems. semi-autonomous systems you might use to OCR while you go off to

[4:13] answer emails. There are systems like Lou used to organize this conference. A few other perspectives that are framing my takes here. AI is a normal technology. This is a great Princeton computer scientist developing this framework. Its benefits and harms are going to look a lot like computers themselves, not hypothetical future super intelligences. So I'm kind of coming from this perspective. And my belief is that historians are, and many of them will soon be running agentic systems, is RAs, maybe even co -authors. Obviously, there's solo authors, too, Marnie was talking about this morning. Expertise will get cheaper in this period, but judgment and taste will become more important. Okay, so let me give a few cases in the realm of administration. So syllabus and teaching management. Here's an example of a very nicely done Canvas website that an agentic system created for me using all of my readings, a syllabus that was written for the class, and hyperlinked nicely throughout so

[5:16] that students can click through. Okay, okay. Look at my computer. OK, so so I think if you're a graduate student entering the profession, you should be thinking in cases like this of how agentic AI systems can help eliminate time. You're not spending on reading, writing, refining your ideas. Fellowships and research report. Brent and I were just talking about this web scrapers to help you identify sources of funding. Everybody should be thinking about deploying these primary and secondary source software interfaces. You can now use these agentic systems in your software management, your citation management, maybe even primary source management systems to more efficiently tag, maybe even analyze. We'll get to that in a second. Some of the stuff you're working on and the university specific formatting requirements. No one should ever have to fill out an expense report again. Here's an example of an expense report Claude filled out for me just this week for

[6:20] a trip last weekend to Berkeley to talk with computer scientists. Okay, so propositions. AI tools are going to reduce these information asymmetries and professional ambiguities, especially if you encourage your students to experiment with them that might otherwise hamper their early graduate student activities. Agentic tools, I think, are also able to liberate our attention from some of these more drudgery aspects of academic life. and allow us to focus on the big questions, the things that got us into this in the first place. So research. There are dramatically falling search costs right now. Search costs for primary sources, for secondary sources. This is happening from anything from OCRing your sources to scraping for new ones. I pretty firmly believe that if experimenting with these tools can be done appropriately with some sort of ethical framework, within our reach a fleet of RAs. And I think this is probably going to lead to sort of a boom in new micro and macro

[7:21] histories, as Jo Guldi and others here have talked about, maybe a return of the longue durée. Just to give you an example here, this is just one snippet from a web scraping program developed with Claude to pull emails out of the cypherpunks, this particular hacker group whose listserv is entirely preserved online. And then these are my highlights after the agent goes and pulls huge block quotes from the sources that are verified by another LLM. And then this is great. This is saving me. This is collecting and collating a ton of information that it would take me hours to do otherwise. So what might this lead to? I think it's actually the advances in agentic systems are actually leading to diminishing returns for being really specialized in advanced computational or digital humanities methods. The easiest entry into this is playing around with agentic systems. I think what we're going to see is probably, in many of our journals and conferences like this, a modern-day methods war.

[8:22] And I think we should welcome it because it, I think, reorients us to first principles. And probably nothing matters more for graduate students than this. Judgment about what questions, what research designs to set your agents out on answering will matter a lot more. reading old books, debating them, getting beers with your cohort mates and coffees with your advisors and advisors at conferences. It's going to matter a lot more. And I think also we're seeing in the social sciences some interesting movements to say that actually AI tools might actually help us militate against our own biases as scholars. So this is a big debate in political science. I'm persuaded that it might be true for history too. So bottom line is that I think agentic AI is going to challenge the historian's research and teaching enterprise might shock the discipline into some needing reforms. For graduate students, I think all of this adds up to a situation where experimenting with these tools, in addition to the broad set of information and archives and questions they make accessible to you, this means there's really no more interesting a time to be studying history, despite pressures

[9:26] on the job market, despite various political headwinds that give us all pause. Because the questions we're asking with these tools and talks in this conference have already pushed us to consider them. What is history education actually for? Who is historical work for? To what ends are we actually working? And I think we're beginning to ask these questions in the way that mathematicians are asking them after the big OpenAI Millennium Prize debate. So this is Ben Recht, an amazing mathematician and computer engineer. I'm writing on Substack a paper dump. And he's saying that AI can now write all these amazing, but probably very exquisite and useless theory papers. But perhaps in seeing how well these models are doing the basic sort of work of research and production we've all specialized to do, we can actually look up and realize we should consider doing something else. And there's a big provocation there, I think, reorienting us to what is the civic purpose of history? What is the political purpose of history?

[10:27] What's the sort of intellectual cultural mandate that brought us all to these questions in the first place? And that, to me, gets me pretty jazzed up about students entering this space. So I'll end there. A few provocations, hopefully, and look forward to the discussion. Yeah. All right. Hi, everyone. My name is Mandy Regan. I am an assistant professor at Clemson University, where I run the new digital history PhD program there. So today I just want to talk a little bit about that program, some of the unique things we're doing, what makes us different than a traditional PhD in history. And I'll leave some things that we can flesh out more in Q&A.

[11:31] So a little bit of context. Our program began in the fall of 2022. Is it? Oh. That's weird. Oh. There we go. No worries. I might have pressed something. So our program began in the fall of 2022. We're just entering our fifth year. Our first students will graduate in the spring. And our degree is different in that it is specifically in digital history. So when you graduate, you have a PhD in digital history, not just in history. Um, uh, and, uh, uh, so, so, you know, the program's sort of motto that I've forced upon it is that the history drives the digital, the digital doesn't drive the history, right? So everything we do is in service of asking and answering historical questions rather than, uh, uh, uh, making visualizations just because they're

[12:35] cool. They have to have an argument behind them. Um, uh, and so, you know, as the, as the director of this program, I also teach, uh, many of the methods courses and, uh, really I've had a front row seat to just how much this field has changed in five years. It's sort of crazy to look back on the first iteration of the methods course that I taught where I made students code everything by hand. And we spent something like 200 hours in my office troubleshooting, you know, where the parentheses missing. And that's why your code is not working. And it has just changed so dramatically. So let me talk a little bit about sort of the way that the program is structured so that you guys have a sense of what we're doing. Um, so in most history PhD programs, right, you have some sort of primary field, U S world, Europe, maybe something more specific. Um, and then you have a set of concentration or, um, uh, minor fields. Every program calls them something slightly different, defines them slightly

[13:37] differently. Um, in our program, you have a primary and a concentration field, uh, uh, as well as a set of core classes like historiography. So all of the core, uh, sort of historical, uh, pieces are there, but you also have a field specifically in digital history. And depending on how you want to count the credits here, right, it's actually more credits than your primary field. Within that field, you take two digital methods courses, which are essentially coding classes. You have a theories of digital analysis class. This class is essentially digital historiography. So while you're learning the methods in my class, alongside of that, you're reading about digital history projects that have use those methods, you're reading the literature, the theory behind that method, as well as getting sort of a brief history of the field. We also do quite a bit of ethics in that class, those kinds of things. We have two rotating topics in digital history that change depending on who's teaching them. We've done things on local history and public archives where

[14:40] students digitized sources from the special collections. We've had GIS and spatial history classes. This semester, we're running a grant writing class that's focused on the American Revolution, where students are writing America 250 grants, while also reading the historiography of the American Revolution. And then the final sort of capstone class in this section is the seminar in digital history, where students build a digital history project of their own that ideally is a piece of, will become a piece of their dissertation. And every student does a digital dissertation in some shape, depending on their career goals. We also, when my colleague Josh Catalano developed this program, this was sort of his baby, he did a survey of all of the PhD programs in the country and sort of asked, you know, like, what did you not have that you would like to have? And the answer from many people was an internship.

[15:45] an internship in, and that's been very successful. We placed one of our first students in a, she's got a great job with a nonprofit consulting company, or it's a consulting company that works with nonprofits. And she got that because of her internship. And so we've had really great success with that so far. And then we have something that we call the Digital History Salon, which is sort of a professional development course, which meets a couple times a semester. Increasingly, I use that to talk about AI because it's just it changes every six months dramatically. And what I taught students three years ago in the methods class is now out of date. So we sort of use that as many times a rapid response. So when we started this program, you know, ChatGPT had just released the coding tools were just, you know, sort of tab completion. And now I teach the methods class pretty much entirely in Claude Code.

[16:47] But one of the things that I fundamentally believe and is a hill that I will die on is that students have to learn to do it the hard way. There are the data that you use, the choices that you make. These are all arguments embedded within the code. And so it's very easy to let Claude Code go and it will write you a ton of code. code does, that's a problem. And so my students have to learn to code from scratch for the first half of the first methods class. They have to learn, we do it in the programming language R because I think it's a little bit easier to learn. But by the second iteration of that course, we have jumped to Python and are doing much more advanced things. But the rule is that students have to be able to explain their code. I will pull up a script and say, what does this do? You've got to be able to explain it to me. And so that's one of the core pieces of our program is that students understand their code. The other thing we're doing that I think is different is,

[17:52] you know, we've really sort of iterated on the comps structure. And it's changed a couple of times. So we adopted oral comps for the primary field and a written exam for the concentration field. us a long time to figure out what does a comp in digital history look like? And how are we going to administer that? We started by just having, you know, sort of two essays that people wrote as take home, you know, historiography essays. We decided that didn't really capture what we were trying to test students on, right? Which is not just a knowledge of the field, but an ability to apply digital methods to historical sources. And so then we developed sort of a technical exam, right, where students had to write a loop, write a function, build a topic model, right, build a word-embedded model, do some of these core methodologies.

[18:53] But we decided that it wasn't 100% testing what we wanted to but it doesn't matter if you can't extract meaning from that. So what we've developed is sort of a hybrid model that's run entirely through GitHub, where students get an exam with primary sources attached to it, and they are walked through a set of three different sections where they have to apply digital methodologies to those sources and then write essays that integrate sort of the history of the field into their answers. and we've done four or five of these now and I think it's working really well and it's a better measure of whether they're actually learning what we want them to learn which is how to make arguments with digital methods, right? I'm not looking necessarily for whether or not you knew where the parenthesis went. I'm looking for you to actually make meaning out of that.

[19:54] So that's where I'll wrap up but we also have an AI policy, which I'm happy to talk about in the Q&A and share with anyone that's interested. It sort of tries to bridge this gap between you can't write your dissertation with AI, but we know you're going to use AI to build your code in some way, shape, or form, and tries to sort of walk that line in doing that. So, yeah, thank you very much. Thanks so much. My name is Chris Phillips. I am a professor in Carnegie Mellon's Department of History. there. The Department of History at Carnegie Mellon is an odd beast. It's long prided itself on not being known for what the entire rest of the

[20:56] university is known for, which is technical and dramatic innovation, and dramatic in the School of Fine Arts, not in the School of Computer Science, although they are also drama queens in their way. But rather, you know, the department has prided itself on being a kind of old-fashioned, coverage-based social history and global history focused department. And so when I took over, it was a very odd sort of place. It was a place that really prided itself on its separateness from the technical parts of the university. I think many of us will work in history departments that have this kind of instinctual you do you and will do us kind of orientation. And many of our colleagues, of course, have the same orientation. And so as I was thinking about both the question for today and more broadly, you know, how to transform a history nowhere and in fact are themselves really interested in digging in in the kind of methodologies that have worked for them. And the fact that many of us in education

[22:00] in general recapitulates training, right? We tend to do the things we were trained in. And it's very, very hard mid-career and there's fewer and fewer resources mid -career to pivot. And so why should we expect all our colleagues to kind of stop on a dime and change how they're doing things? And so in that kind of environment, I think the question of graduate education is a really tricky one, right? You have a group of people who are ultimately responsible for attracting graduate students, for training graduate students, for advising graduate students, who themselves do not have the time, inclination, or resources to pivot to the kinds of novel programs that we just heard about. And so how do you think about graduate education in that particular setting? I will say my own orientation, I'm trained as a historian of mathematics, statistics, data quantification, that sort of thing. And I feel as if all of us, you know, in the sciences, they used to say you're either a lumper or a splitter. I don't know if you guys remember that. Either you see everything is the same or you see everything is different.

[23:00] I think in history, we're all either continuity people or discontinuity people. And I will just throw it out there. I am a continuity person. And so when I see the question of digital methods or computation, I say this is something we've been dealing with for a long time. And the question is not how do we reinvent ourselves, but the question is rather how do we think about the pivot, if you want to use that kind of language, in this long tradition of what we've been doing well. And so one of the things I think that all graduate students should think about is that historians already use computation. The most fuddy-duddy professor is looking up everything on computers, using AI-driven search models, using the kind of tools that are built into JSTOR, even though they imagine JSTOR to be some sort of library-like equivalent. And so we need to kind of foreground all the assumptions that go into the way in which normal, so-called old-fashioned search work. That is to say, everyone already does computational history.

[24:05] And I think normalizing computational history in that way can make the kinds of questions that the new technologies enable much more familiar to those of us who've been doing this work for a long time. More to the point, I think, pretending like this is new and that you can avoid the new by continuing what you're doing actually increases the kind of blind spots we have. Laura very famously wrote on this about a decade ago in the AHR. Many of us are familiar with that seminal article on the kind of blind spots that come about because we're not intentional with the ways in which we engage with digital resources. I think that kind of training is still the essential training in our age of LLMs. That is to say, you are making choices about how you analyze data. Those choices are going to create blind spots. They're going to create opportunities. you can do things the way we've always done, or you can tool up for our modern 21st century history department. It's rather you've made a choice, and that choice has

[25:06] costs. And if part of that cost is that you can't look at many documents, as you might be able to look at, if part of that cost is you're stuck with the kinds of methods that your advisor used, then we should think about those in terms of costs and benefits, instead of it being a kind of, well, you can either take the time to learn these skills, or you can just do it the old fashion way. I think that's the wrong kind of way to transition a graduate program or to transition a faculty. In terms of the practical outcomes of this, I think we should stop thinking of digital methods as special topics. It's kind of like those sections of the bookstore where you're like, why is this in its own section? This is just fiction. This is just nonfiction. Like, why is this the special section? And I think we all know the special section is there because it needs special protection. It needs somehow to be highlighted as different, as odd. And I think we should all just get on with the fact that every methodological choice we make has benefits and problems. They all come with their own drawbacks. And I think that language we use in

[26:09] graduate training and more broadly in the profession would do us all a favor, even as we acknowledge that AI has really changed the way in which many of these computational methods work. And so I think things like database management, data analysis, search retrieval. Instead of these being special courses, we need to rethink the courses we already do in terms of how do you use archives? What do you do with those 10,000 JPEGs you took because you only had four days in the faraway place to go to? That's actually a data management, a computational question. And I think we should introduce that language in those kinds of classes just to make it transparent that there is different ways you can actually manage that kind of aspect of the graduate training. I will also say, in my quasi-controversial way, we live in a world in which by the time graduate students apply to history degrees, they have opted out of every possible opportunity to learn technical

[27:13] skills. The pool of people we're drawing from in history is probably the least technically competent pool of people, maybe in the entire university, because even the English department is doing better because they've had long kind of text analysis and kind of text heavy processes. And so I think we should be open about this fact that these are folks who've said no to the tech train over and over and over again. And so then to turn around and be like, and now in your first semester, let's talk about database management. You know, they're like, I don't know where files are on my computer. I don't know what a terminal is, except if it's a train. You know, this kind of, we should just be honest about that. And I think it's okay. It gives us an opportunity to rebrand it. But I think that's all the more reason why we have to normalize the practices we've been doing for decades to make clear that those are already computational practices. Computers have been used by historians for a long time. And that it's not that this is all a new thing.

[28:14] It's rather that these are new tools, the same way Microsoft Word or the same way in which you can really see I'm a continuity person, the same way in which the kind of search engines were new tools. The second thing I'd say is that, and this is partly about graduate training, but also I think as a chair, this is what we should say to our deans. History is a social science. It's always been right there at that humanities and social science benchmark. And I think we should start thinking of history again, apologies if this rubs you on the wrong way, as an end of one experiment. In fact, like economics has it easy. They can just keep running the darn experiment over and over and over and over and generating data. We can't do that. We have an end of one experiment and we have to make inferences about the past on the basis of this record that we've been given. And so every choice we make, what's our population? What's our sample? What kind of inferences are we making about the past? regularities are we assuming based on the data we've seen? Are we making assumptions because we only read English?

[29:16] Are we making assumptions because we only look at what's been printed in journals? Those are all statistical assumptions about regularity. And the more we can start to talk about them as assumptions that we are making about how to make inferences, causal or otherwise about the past, I think the easier it is to bring people into our kind of modern world. And so I think in a practical sense, we should embrace history as not just making judgments about the past. That's critical. As multiple people have said today, we need to continue making judgments. That's what we're good at. But also embrace our grounding as a kind of social science. And I think, again, that makes our training less distinctive and less problematic in our kind of age of everyone looking to see what can be pulled from the modern university. And as a practical matter, I think our students should be encouraged to take classes in other departments that are also dealing with problems of causation, inference, and interpretation. We're used to people taking classes in the other

[30:18] humanities. I think we should open our minds to really think about taking classes the way other folks think about causation or inference. And therefore, we know what historians do particularly. And the final thing I'd say is that the people we're training now, this shocks me and I'm not that old, but the people we're training now are going to be in their career prime in the 2050s. And so, like, that's the deal, right? These are the people who are showing up to do graduate training now. And that means the 1980s is distant past. these folks in the prime of their career are going to have to deal with a huge swath of history that has born digital material, that has scanned material right cheek by jowl with synthetically produced material, that has to deal with proprietary file systems, that has to deal with like the way you and I document our own lives, whether that's through our iPhone, whether that's through our email, whatever it is, these are the problems of the future.

[31:18] And so graduate training, the idea that we can just say, well, I hope you pick a topic pre-1970 that involves languages that still have archives, it's a kind of disservice to the actual future of the profession. And for those of us who are privileged enough to work in institutions that still have graduate training programs, it is our obligation to be producing students who are able to take on these challenging roles of the future. And so I think in a practical sense, we need to think about how we approach these new forms of evidence with a historical mindset, with our own expertise. And then the last thing I'd say, just to kind of close on, is that I think the questions that we, me, I spend a lot of time worrying about, like, what's our policy on AI for dissertations? What's our, what kind of class? How do I get someone to teach that class? These questions should come from the sorts of big, major ways in which this question of digital methods is a question that we have been dealing with for a long time.

[32:21] And I think the answers then are less about how do we adapt to this kind of brand new world, and rather, what have we always done as historians, and what does that look like, given the context that we face today? I'll stop there. Do you have slides or should I? Should make. There we are. All right, this is a rarity for me to not have slides because I'm a historian who works with material culture and yet I don't have slides. Hi, everyone. I'm Sarah Weicksel. I'm the executive director of the American Historical Association. And my job is not just to think about the future of graduate studies but to think about the future of the discipline itself. And to lead the AHA in conversations about that.

[33:21] And that, of course, depends a lot on how we are imagining the future of doctoral education. So my comments today are going to be more around the work that we're currently doing at the AHA to think about these questions and to identify a couple of questions that I have that I'm trying to work through and that I invite you to help me work through over the years. So one of the things that we have that is ongoing right now is a project with the American Council of Learned Societies, ACLS, that is called Doctoral Futures. And we're doing that in collaboration with the Modern Language Association and the Society for Biblical Literature. And really what that project is about is a question about what should we, do we need to be doing right now to ensure a vibrant future for the

[34:22] humanities in 5, 10, 25 years? So we've been working for a year with three committees, each one led by one of the societies, to assess all of the work that has been done over the past 15 years on graduate education and on career diversity initiatives, like the one that the American Historical Association ran in the 2010s. What have we learned from that? Where do we fail in those efforts? How do we reconfigure them in a way that is actually helpful and useful? How do we offer departments opportunities to expand the way that they're thinking about doctoral education in ways that make sense for their institution? How do we create essentially a menu of options for people to

[35:22] think about how they can transform some of the aspects of their programs, keep what's working really well, and reconsider what might not be working so well? That has been a really eye-opening experience. One of the things that has become most clear is that for history students in particular, but also those working on PhDs in other humanities disciplines, they are coming to graduate programs with the desire to go into a wide range of careers. and yet we are continuing to because it's very comfortable because we are used to it because it's how we were trained we're continuing in many instances to uh think on a think that the the type of doctoral education the types of requirements the types of courses that we've always required can

[36:25] easily transfer over to a new a student's newer idea of what they might be able to do and people at this conference today who are working in a lot of different contexts that are not academic departments. And I think that that is also very telling as well. So how do we ensure that those students are not only able to develop that deep knowledge of the discipline that is so essential, that is what actually allows us to be better at being interdisciplinary, allows us to be better in the kinds of employment we go into? How do we develop that deep, deep knowledge of a discipline, but also gain opportunities along the way to try out different things? How do we gain different skills? How do we have internships that allow people just to dip their toe into something in a way that allows them to think more expansively

[37:25] about what their future might hold? Because sometimes a student enters program assuming that they want to do one thing and then wants to pivot but doesn't see an opportunity for how to pivot. So how do we help departments prepare themselves and their students for those pivot moments? One of the ways that we've been thinking about this at the AHA is around the idea of broadening the definition of historical scholarship. That we have for, you know, decades, and AHA has been around since 1880, so over a century, valued certain kinds of products that historians create more so than we have valued others. We value books. We value articles. We value more than we once did the digital projects that so many amazing historians are producing. But there are so many other types of scholarship that

[38:27] exist. And scholarship is not always about producing new knowledge. It's also about diffusing knowledge. It's about bringing your work not only to an audience through a book or an article, but also about an op-ed or a public lecture or a workshop that you're running with a local high school or a digital university. exhibition or a set of data that people are able to use. There are so many different kinds of ways that we are making our scholarship more accessible to not only other scholars, but also to the public. How do we value that in our history departments through primarily through tenure and promotion, but not just through tenure and promotion. How do we value that and celebrate that as people and as colleagues? Because it's one thing, I think, for it to be on the list of things that you're

[39:29] able to do. But if there is an attitude that these other things are somehow lesser than, that continues to create a culture in which it seems less acceptable to step outside the box and explore how to make your scholarship more accessible to different groups of people. And it also, when you have a environment in which faculty are celebrated for all of the ways that they are doing their work, you also have graduate students who are able to better imagine what they would like to be able to do. And it's easier for them to ask the question, how can I, where could I, what do I need to do to be able to do this other thing differently? So some of the other questions that I'm really thinking through as I travel around the country, I visit departments, or how do we address the fact that

[40:34] PhD cohorts are shrinking? And in that context, how do we ensure that students are able to have the critical conversations that you need to have to develop that deep knowledge of history, of how do you develop the subject expertise? How do you have the historiographical debates when you're only one or two people in the same cohort? How do we support departments that are seeing funding drastically reduced? not just from federal funds diminishing or going away altogether, but also a reassessment of where resources go within a university. How do we ensure that history is able to take full advantage of what is available to make demands about having more resources, but also

[41:36] as we're working on that larger problem of getting more resources for our history departments, how do we help people to work within those limited resources? And then what about the lone digital historian? I'm going to bring it back to some of the topics we have here. Someone who would like to be moving and doing one of these pivots, who wants to have access to additional, but needs more resources, they need more storage space, whatever the case may be. How do we help connect people across institutions, across departments? How do we help people to think less about the kind of we work within the box of the university and instead that there is a great deal of value to going outside, creating these kinds

[42:37] of partnerships? That's a much larger institutional problem that we're trying to solve. So in terms of AI, we have two committees that have been at work and are continuing their work. We had a committee on AI and teaching in the historical discipline that wrapped up their work. And we will be reinstating a new committee who will be working on new questions. And I suspect that graduate education is going to be one of the topics that will be at top of mind for that committee. And we have another committee that, and there are a few members of that committee who are here this week, who are delving into questions of research and also publication. What do we need to know? So I'm actually going to kick us off. know, what we're doing

[43:38] in terms of thinking about graduate education, but also wanting, you know, really wanting to ask the question, part of the reason that we wanted to co-host this conference, what are the opportunities that AI presents us with? What are the problems it creates? How are historians uniquely positioned to take advantage of and to solve both of those things? So the question I wanted to kick us off with to then throw it back all to you is, for our panelists, what is the one question that you think all departments should be asking themselves right now? What's the question that next month needs to be on the department meeting agenda? I will walk slowly back to my chair.

[44:52] I'm the chair. Oh, geez. Yeah. I don't, I don't, you know, we all know the chair doesn't set the agenda of these meetings. We just try and herd things toward an amicable resolution. But I think one question that all history departments will be asking, although they won't actually ask it in exactly this way, is what is the specific intellectual contribution of history departments in the constellation of other things the university is doing? Because it's not just history that feels threatened methodologically and pedagogically by these new technologies. Computer science is freaking out way more than the history department in terms of the overall impact of the technologies on the pedagogical and research missions. But I think in that context, it creates an opening where departments can be clearer about the relevance of their methodological commitments, their foundational values, etc.

[45:55] And so it's actually, it's a rare time, I think in the next year or two, where all departments are going to have to do this. And so instead of it being a crisis of the humanities, where, you know, two thirds of the departments get to just continue on status quo ante, and just a small group are called out for being problematic. all the departments now are at a moment of opening. And so I think all departments have this opportunity. So I think for a history department, we should say, just like you guys are thinking about this question, here's what we're doing and here's the specific value we have. But I don't think we're having that conversation as widely as we might. Yeah, I agree. I think we need to find a way to better communicate the value of a history degree or a humanities degree to our students. And I think there's actually an opening here. I keep reading about AI CEOs saying that if they're going to send their kids to college, they're going to send them to a liberal arts school. We teach students how to think critically. And that is the core skill you need to use AI

[47:00] effectively. I always tell my undergrads, you know, you don't pay a robot to go to the gym and lift weights for you. why would you let ChatGPT write your essays for you? Right. We teach you how to think and then AI is more useful. And we have to figure out a better way to communicate that to our publics. So I've been to four faculty meetings. So this is, you know, this will be colored by that. Exactly. It's a blessing, but I've also been to four now. committee meetings with the AI fluency initiative at Ohio State. So unlike many other universities, Ohio State announced this plan to have every faculty member and student fluent in AI use by like 2029. This means that as a sort of enterprising young historian of computing dropped into one of these interdisciplinary centers, I'm immediately pulled onto this committee for my sins. But one of the questions that we're posing to the departments from

[48:04] that university level committee and from the Interdisciplinary Civic Center, where this is my tenure home at Ohio State, I'm affiliated with the History Department too. One of the questions we're asking is, how do you govern an agentic, specifically these LLM tools that our students are using and researchers are using to flood journals, to check bibliographies, to tag and look at vast corporate of sources like Lou and Sam Backer are doing and their project in the B&O Railroad. How do you govern that use in your professional life? Right. So I think it's what is acceptable for research or not? Why? And I think you have to be able to articulate that. And it stems directly from this question of what is the value add? Right. And I think there's a lot of thorny questions about the sort of publish or perish paradigm. Like we don't want to end up in a situation where historians or historically minded, you know, interdisciplinary scholars are doing what the University of Chicago statisticians doing, which is pumping 200 pages, 200 papers already as of a month ago on arXiv and all the various journals he's

[49:08] publishing. That's a disaster. So I think we, you know, we ask, how do we govern this, these tools in research and teaching in our units? What best practices should we adopt? and then kind of from that question it'll get us to some of these other purpose questions as well. I'm going to open it up to the floor. See how I dodged my own question? We're a full function faculty here. Okay. Please introduce your name and speak your question to the microphone clearly. Yes, I promise. from Carnegie Mellon. I'm one of Chris's minions. And I'm really kind of curious about if we are training history PhD students who are not necessarily going to become historians, which I think is important, you know, are we asking them, is the

[50:09] setting the bar too high if we are also kind of maintaining this myth that the historian is a lone scholar? So I've heard a people say, well, with AI, we don't need to collaborate anymore. We don't need the computer scientists. Now, I completely disagree with this. We don't need the data viz expert. We don't need the NLP person. But, you know, the training in those fields is a kind of apprenticeship training where you write papers, you co-author papers with your advisors and other colleagues in order to learn the methods and, you know, learn these practices, which are valuable both within and So are we asking too much of our students if we continue to maintain this idea of sole authored work as like the pinnacle of historical scholarship in a context of computational methods?

[51:14] yeah i think um i don't know that i have a a straight answer to your question but i you know i keep thinking about how um you know i went to george mason university um and i worked at the Rosenzweig Center when I was there. And so I got a lot of hands-on project experience, project management, things that you can't really replicate as well in a classroom. And that was incredibly valuable to me. And one of my fears is, you know, I just started a new project where we're taking lesbian guidebooks and digitizing them. We're using, we've built an AI app that transcribes them. And we were able to transcribe, you know, 16,000 locations in a period of like two weeks. And that was with us checking every page. My previous project, which used 30 years of guidebooks, took us six years and we had an army of graduate students working with us and we trained them

[52:17] extensively. What happens to graduate training in these digital humanities centers, in these places where collaboration when all of a sudden it's cheaper and easier and faster. And in many cases, it's better transcription because humans make errors. What happens to that training model? You know, I think it's something we're going to have to think a lot about because that collaboration, that model was so valuable for me and I know has been so valuable for others. This is one of the key questions that we're working on in the doctoral futures program. And of course, it's not specific to history, but the question of whether or not the dissertation as we know it is or should continue to be the thing that we require for

[53:20] the PhD. and if there should be different tracks that students select in terms of the kind of end product. What is the skill that we are actually trying to assess when we have students do the deep research and the writing to produce a dissertation? I've been in multiple conversations, convenings around the question of collaborative research in the humanities, how it can be done and, you know, questions that expand from there. In addition, so in addition to what Chris had said about how, you know, we're all, we started off as the people who were, you know, eschewing technology. I think that a lot of historians were also the kids in high school who were like, I do not like the group project because I ended up doing it all. And so it takes some effort to break out of that. and to one of the things that I have always taught my

[54:24] students is that even though they think when they are walking in the door that they are there to do the thing that they want to do and they're going to do it alone because they're going to go sit in an archive alone they're going to do the research alone they're not and at the basis basic level we are all engaging with historiography. You can't do history alone. You have to be in conversation with other historians. What happens when you take that conversation and you turn it live and you don't do it just through the books and to really think about what that kind of collaboration looks like? So I think that in the end, what we really do need to, and this is a department kind of level decision. And I think it is one of the questions departments should be asking themselves right now about what are the requirements for the degree and how do we ensure that we are sending our students off with the skills and the

[55:25] the skills and also the the confidence that they know how to be a historian in the world in multiple ways yeah it's even worse than that in some ways because the uh the you know the questions i always get from the dean is you know why do you have graduate students if they don't produce research that has your name as a senior author on it you know what what is the value of graduate education if it doesn't further professors research agenda because in every other field graduates do a lot of the legwork. You know, you have a pyramid of faculty, postdoc, advanced graduate students, undergraduates, all the way down that funnels through. We've known this for decades that this is how it works. And so, you know, I think on one hand, I think recognizing that history has always been capacious, whether it's people who co-wrote but uncredited wives for decades as researchers that we didn't, you know, going back for many, many decades,

[56:32] long and inglorious tradition, all the way to the fact that we deeply advise graduate student work, but never put our name on it. And so all of a sudden, there's a kind of invisibility there to then we turn around and tell the graduate student, now you're doing it on your own, and you're having to do it. So I think it's, I guess what I'm trying to say is at every level, of the scholar all the way through to the justification to the dean of the existence of a graduate program, there's a kind of oddity of historians. In many ways, we're one of the last remaining holdouts of the single author being the default. And even English departments, I think, over the last 20 years, MLA has really done a long, long job pushing for new ways of giving credit to co-authorship for promotion and tenure that we haven't done. But I think it's a good question. I just see it as part of like, it's just a huge, in a way, the computational role

[57:33] here is almost the uncredited computational work that I think is where does that fit into the broader problem of the uncredited work that historians do all the time. And I think in a way we should deal with this bigger issue. And then the specific role of how do you deal with software engineers or how do you deal with data viz people will play itself out. And Louis is giving me the signal, the universal signal to wrap it up.