Author: Mark Hopwood

Leslie Todd, Art History

Students play the Prado Museum Expansion game in ARTH 111

In Spring of 2022, I presented in a “Teach and Tell” at the Center for Teaching with Amy Patterson on using “Reacting to the Past” games in the classroom. Reacting to the Past (RTTP) is: “… an active learning pedagogy of role-playing games designed for higher education. In Reacting to the Past games, students are assigned character roles with specific goals and must communicate, collaborate, and compete effectively to advance their objectives. Reacting promotes engagement with big ideas and improves intellectual and academic skills.”

I was first introduced to RTTP as a graduate teaching assistant for a humanities course at the University of Florida. Because we played two games in the course and I was new to the pedagogy, the professor for whom I was working generously applied for a grant to include me with her at the RTTP consortium’s Winter Conference. While there I learned more about the technique and participated in playing in an abbreviated version of a game to gain firsthand experience of what we were asking students to do. By the end of the weekend, I was fully invested in the benefits of the games and was energized by its possibilities. As I expressed at the 2022 Teach and Tell, it fosters student autonomy, active engagement, and experiential learning. It engages students’ use of primary sources while exercising their speaking and writing skills and pushes students to inhabit alternative perspectives. I was most enthused by the possibilities of experiential learning that the games offered since, coming from a humanities discipline, the options tend to be restricted to field trips or other one-time only types of activities all of which can be costly or are tied to other limitations. Classroom gamification was an appealing model.

In Spring 2020, I finally had the opportunity to implement an RTTP game in my own classroom. I teach ARTH 111, Introduction to Latin American Art every spring semester. I organize the course chronologically with the first third addressing the ancient period, the middle third on the early modern period, and the final third on modern and contemporary art. From the beginning, I used RTTP to teach modern and contemporary art through a game entitled “The Prado Museum Expansion: Diverse Art of Latin America.” While most games ask students to reenact historical events such as the French Revolution, for example, this game is slightly different because it is based on a fictional scenario. In 2010, the Prado underwent major renovations. The author of this game imagined a scenario in which a part of that expansion included a new wing to incorporate Latin American art of the twentieth and twenty-first century. In the game, most students were assigned the role of living artists or art dealers representing deceased artists all vying for a limited number of spots available in the fictional new gallery at the Prado. The remaining five roles were of leadership positions within the museum who ran the game and determined the final winners. While the objectives of the game are straightforward, each student is given their own secret role sheet with gambits and somewhat underhanded motives that keep things interesting. 

I played the game the first three years I taught the course. Students were energized and enjoyed it once they got into the new rhythm and pace, often responding with positive feedback at the end of the semester. The most common responses were that the game was fun, exciting, engaging, different, and unique. It was rewarding and invigorating to see students gain the confidence to take control of the classroom while exploring broad thematic questions such as “What is Latin American art?”  After each semester, I would assess areas of improvement and make changes for the next year. However, I always felt something was slightly off or missing. During game debriefing, it became consistently clear that students were more focused on the intricacies of how the museum world operated. Understanding modern and contemporary art styles alongside the cultural phenomena that informed them became secondary. I finally had to confront the fact that as fun and useful as the game was and as much as the students were gaining from it, it was ultimately not fulfilling my primary course objectives in the way I originally envisioned.

In Spring 2024, I finally stepped away from the game and tried something new. I used the game role playing, participation activities, and writing and speaking assignments to formulate a new, end-of-semester assignment. I asked students to choose an artist from a list (following the list of artists from different art movements represented in the original game) and to take on the “role” of “teacher” to give a brief lecture explaining their chosen artist’s background and art style. They then had to write a paper comparing their artist with that of another student’s artist—ensuring they were still coming to class and paying attention during student presentations. Once the presentations were complete, we addressed the common themes that arose across the artists and assessed the question of “What is Latin American art?” While not a perfect reflection of the game, I took aspects of the game as inspiration to offer students a similar version of that same autonomy while remaining more aligned with my course objectives. To my surprise, I received very positive feedback on the assignment. Students like to hear from one another.

I am still a major proponent of RTTP but recognize that not only can it be intimidating for a non-RTTP inductee, but not all games perfectly align with our courses. If the idea of using three weeks of the semester to play a game in a singular topic feels like too much of a sacrifice to your course content, I suggest ordering a teacher’s copy of a game to pull ideas and formulate an activity and/or assignment drawn from an RTTP game. We don’t need to reinvent the wheel—just poach and transform the wheel. If RTTP is something you’re interested in, they do offer a limited number of micro-games on certain topics that are playable in one class session, which could be used as a model to invent your own micro-game!

Jennifer Matthews, Theatre

After spending most of my time at Sewanee solely as a costume designer, it seemed time to ‘brush up’ on my dormant scenic painting skills to be able to teach scenic painting in Advent 2024. As a graduate student at the University of North Carolina at Greensboro, our scenography program prepared me to become a diverse theatre artist- one who thinks of the entire design in a holistic way, and many of us had significant second areas of design.

Although costume design will always be what I am most drawn to, I appreciate the process of scenic design, and as part of my training, I spent a year as a ‘scenic charge’ during my graduate assistantship. A scenic charge transforms the scenic designer’s work from paper to paint. This position requires mastery of a variety of techniques to paint textures of all types, as per the desire of the designer.

I was fortunate to be able to bring Jennifer Rose Ivey, author of The Scenic Charge Artist’s Handbook to the Tennessee Williams Center in July for an intensive workshop in traditional and new techniques for scenic painting. Additional focus on material safety and best practices provided insights that encourage creativity while focusing on well-being. Ms. Ivey is a gifted and imaginative teacher who consulted with me on current pedagogy of teaching scenic painting to reimagine THTR 346 to current professional and creative standards. While in residence, Ms. Ivey performed an assessment on our scenic painting studio and resources.

When attending the Theatre Sewanee performance of Natasha, Pierre, and the Great Comet of 1812, the community will get to see our collaboration. Ms. Ivey is the guest scenic designer for the show, and my students in THTR 347 and I will be painting her designs on the set.

Chris Shelley, Neuroscience

Following our successful application for a Department and Program Enhancement Grant from the CfT, the primary Neuroscience faculty of Chris Shelley, Kate Cammack, and Thomas Reppert met from 9 am to 4 pm on May 21st. The aim of the meeting was to redesign the neuroscience comprehensive exam and evaluate our major assessment data. The main reasons for the comp redesign that the current comp is based on our 200-level course and we have recently restructured the major (reducing the number of 2XX course from three to two), and the increasing abilities of AI to answer our comp-style questions that involve analyzing primary research literature.

We began by examining examples that highlighted the need to update the comp format in light of the growing abilities of AI. We examined the responses that the website ChatPDF gave to all of our comp exam questions. This is a website which allows you to easily upload a pdf of a research article and then receive AI-generated answers to questions about the article. Although we all agreed that the answers were not perfect, in many cases they would have been of sufficient quality to pass the pass/fail comprehensive exam, especially if students would have had time to edit and modify their responses as coursework.

The neuroscience comp has evolved over time, in part due to the logistics of increasing numbers of majors (2019 had two majors, 2024 had eighteen majors) and in part by trying out different formats and timings. We reviewed how the neuroscience comp had changed since the major was introduced in 2019. In the first comp, students had to write a mock grant proposal. In later comps students needed to analyze tow primary research papers out of a choice of three (based on the three 200-level course) and then answer short essay-style questions in the first Easter comp. period either in exam conditions (2020 and 2021) or as coursework (2022 – 2024). The questions were given to the students either in the exam (2020), at the start of the first Easter comp. period (2021), or during the last Advent comp. period (2022 – 2024). We agreed that in light of AI and the restructured major that changes have to be made to the comp.

We next reviewed the goals of the comps as detailed in the CfT application, our major’s learning objectives, and the rubric for the 2024 comps. One thing that we felt was not being emphasized enough in the comps, and is a key goal, is the connections we would like students to make with material across NEUR courses and their own personal scientific narrative throughout their major. We then reviewed the comps exam format in Biology, Psychology, and Biochemistry and the capstone course guidelines of the Environment and Sustainability major and the College of Wooster’s capstone research course. Across these areas, tools such as multiple-choice questions (MCQs), analysis of research papers, short answer questions, essays, and mock grant proposals are all used. We also realized that we did not have any information on the comps in either Chemistry or Computer Science (majors with close overlap with Neuroscience) and so will find out how these programs structure their comps.

We spent the rest of the meeting brainstorming potential comprehensive exam formats and methods of delivery, and linking these to our major’s learning objectives (below):

• Personal statement / articulation of interests
• Basic Neuro content, Qs
• Being able to articulate why specific data is relevant to neuroscience/society
• Providing an explicit way to connect ideas across several courses in the neuroscience curriculum / integration of concepts from multiple sub-disciplines
• Describe and explain one piece of content at level of NEUR101, NEUR2XX, NEUR 3XX, NEUR4XX.
• Facility in working with key concepts in scientific papers
• Coming up with their own experiment that’s reasonable/plausible
• Critiquing data. Supporting/refuting hypotheses with data
• Explain a concept at various levels of depth to a non-NEUR major
• Authentic approach – different audiences/formats
• Responding to questions in an oral examination format
• Students write and answer their own comp Qs.
• Collecting a data set and interpreting those data in the context of a developed working hypothesis
• Produce a product that’s relevant to their interests/skills/paths as they move forward
  

After discussing the above ideas, we settled on a new three-part comp that we believe covers the four learning objectives of the major (LO1 Demonstrate knowledge of basic concepts in neuroscience, LO2 Demonstrate the ability to think critically and integratively, LO3 Develop and apply scientific inquiry & research skills, LO4 Develop and apply effective communication skills).

Part I: Basic knowledge test. In person short answer/MCQ basic neuroscience content questions. The content will be based on material from NEUR101 that students will have also revisited in the 200-level NEUR courses. Tests LO1 and LO2.

Part II: Research paper critique. In person exam, in which students will pick one of two papers, and the questions will be less focused on basic knowledge and more focused on experimental rationale, data interpretation, and future directions in terms of future experiments or the societal/scientific importance of the research. Tests LO2 and LO3.

Part III: Metacognitive integration. Pre-recorded video submission with the emphasis on oral content rather than video production skills with no graphics or slides to be used. The content will be based on describing how their major courses contributed to their choice of paper in Part II. Tests LO1 and LO4.

We set ourselves the task of thinking potential logistical, equity and accessibility issues that may arise from the above format, and to develop student guidelines for each part, and example questions, and to have these in place before the start of the Advent 2024 semester.

Anna Foy, English / Humanities

On Sunday, April 28, 2024, members of Humanities 216, “The Classics and the Colonial Americas,” piled into a van outside Gailor Hall and traveled about five miles over bump, dirt roads into an isolated section of the Sewanee Domain now known as “King Farm.”  Large portions of Sewanee’s 13,000 acres are now, or have never been, regularly inhabited by dense human communities.  But King Farm—now itself uninhabited—reveals evidence of human habitation, or at least human presence, dating back some 10,000 years.  Some of this evidence has been retrieved in archeological excavations of the soil, and of a homestead that once stood in a grove of maple trees and poplars.  In addition, on the face of a nearby rock shelter, just above human height, a figure that experts believe was probably painted some 1000 years ago can be seen by those with a trained eye.  (See the photo below, where John Willis has enhanced the image with a color filter.)  We in Humanities 216 wanted to see this ancient image up close, and we wanted to see it in the environmental space in which it had so long persevered.  So we asked Professor John Willis (History) to give us a tour.  So a collection of interested students and their interested instructor (Professor Anna Foy, English and Humanities) spent several hours scrambling over rocks, examining the remnants of an old cow fence and an old well, admiring iron-laden springs, observing graffiti left by generations of Tennesseeans, and paying particular attention to this strange, ancient image, which has become faded enough that we had to squint and scrutinize it to make out its contours.  King Farm, as documented by Professor Willis and other scholarly observers over many years, looks deserted now; but it has been shaped by intermittent generations of sustained human residence, including a period in the 1700s when Europeans ceded the property to the Cherokee by treaty, and also a period in the late 1800s when the property was owned and managed by Dennis Barnes and his family, formerly enslaved, who raised livestock for slaughter and sale after the Civil War.  Our field trip contemplated these layers of human history, with an emphasis on the ethics and logistics of appreciating lasting signs of ancient indigenous life in America.

In doing so, we were continuing a semester-long contemplation of the role of “the ancients” in the education of colonial Europeans, Native Americans, and Africans, and the further question of how we see studying the ancients as part of our education and cultural heritage today.  There is a significant modern history of seeing the study of Greco-Roman writings and culture as the study of “the classics”—i.e., time-tested works of history and literature that are seen as an important component of Western Civilization.  Humanities 216 looked at a starting point of this concept in the late colonial period, when an educated English speaker had to know Latin and some Greek merely to enter college, and when ancient Greco-Roman writings were prominent enough reference points that they served as evidence of the workability of republican government during the writing of the Constitution.  Yet, before the Revolutionary period, most colonial writers did not yet talk of “the classics,” but of “the ancients”; and the category of “the ancients” included a much broader array of nationalities than later notions of “the classics” tended to presume.  In the second half of the course, we investigated this difference in terminology—an interesting case in which the later term (“the classics”) really seems less inclusive or politically progressive than the earlier term (“the ancients”).  Students were asked to do a creative assignment in which they located an ancient (pre-Christian) object or artifact or allusion on campus and then explored the ways that item had been honored, mocked, effaced, or obscured by later interpreters.  Students then posted and discussed their findings, which included some Greco-Roman classics, but also lots of interesting examples beyond those nationalities.  Our trip to the King Farm section of the Domain was, in a sense, a culmination of that experiment, with an emphasis on signs of ancient indigenous life on the Cumberland Plateau that remain somewhat hidden from Sewanee campus life.  

We prepared for the field trip by reading Samson Occom’s petition to the U.S. Congress on behalf of the Brotherton Tribe (1785), whose references to the Hebrew Bible demonstrate an awareness of Occom’s sense that the “dominion” given by God to humankind over animals and plants had been unduly, misleadingly coopted by European colonizers.  We also read a more recent essay called “Unnamed Caves” by a Sewanee alum, John Jeremiah Sullivan, about the thousands of years of indigenous peoples’ activities still evidenced on the Cumberland Plateau and variously known to other human inhabitants of the last 250 years. 

Jennifer Michael (English) and Tiffany Momon (History)

Our joint trip to Mongtomery to the Equal Justice Initiative sites on October 28, 2023 (supported by a CfT mini-grant) was a powerful experience for all. Professor Michael’s class had just finished reading Marilyn Nelson’s crown of sonnets A Wreath for Emmett Till and had written a paper analyzing the use of form and the elegiac tradition in two of the sonnets. Visiting the museum and memorial was a way for them to understand more fully the depth of the legacy of slavery and Jim Crow as well as to consider a physical memorial alongside a literary one.

On the Monday following the trip, Professor Michael’s students wrote a reflection in class in which they were asked the following questions:

1. What single experience or exhibit stood out to you the most? Why?

2. Is there an aspect of your own life that this visit has touched?

3. How has this visit deepened your reading of A Wreath for Emmett Till?

The responses were thoughtful and illuminating. Here are just a few excerpts:

“I think the exhibit with the jars of soil from separate lynching sites really stood out to me because it provided a perspective on the countless jars of soil there were. The fact that when I went around to the other side there were more jars almost made my head spin.”

“One aspect in my life that has been touched is how little information they teach in school about the ongoing acts of racism. In high school and middle school we learned a lot about slavery but we never talked about the ongoing effect it has on current lives.”

“Especially after Nelson’s commentary on what the nature around the site of Till’s death must have ‘seen,’ the soil lasting past the tragedy and being an inconspicuous marker . . . showed how lynchings in the past still affect many people and places today.”

“I was curious if my hometown and other important places to my life were represented in the memorial, and I found every single one of them, including where Sewanee is. I also took time to reflect on how I have long perceived myself as ‘well educated’ on the topic of racial justice, as I found that I learned a lot more from this experience than I expected.”

Students have continued to reflect on the trip. As the final class assignment is to revise one of their earlier essays, several have chosen to revise the essay on Nelson’s Wreath (completed before the trip), bringing their insights from the visit to bear on their reconsideration of the text.

Celeste Ray, Department of Anthropology

Students in discussion with Wigwam Village No. 2 owner Keith Stone

Anthropology students taking the special topics course “Sacred Realms and Biocultural Diversity” traveled to Mammoth Cave in Kentucky this September. The trip concluded a course unit on caves and sacred darkness and laid groundwork for class themes including Environmental Justice and Environmental Jurisprudence by prioritizing place- and issue-based knowledge. In preparation for the overnight fieldtrip, students read multiple articles on Mammoth and two in particular by cultural geographer Dr. Katie Algeo (Western Kentucky University). One considered the making of place at Mammoth Cave and the second examined “The Wigwam Village, No. 2” where we stayed the night. The “Wigwam Villages” were a 1930s motel chain (featuring steel “teepees” rather than wigwams) developed for early automobile-based tourism in 6 states (only one other remains in operation in Arizona). Indigenous peoples never lived in either teepees or wigwams in Cave City, KY. The current owners, practicing architects from New York, view the site as a teaching tool and after an introduction and discussion of their perspective, we were joined by Dr. Algeo for conversation about the motel chain’s almost one century-long history and the evolving structures of power and privilege that make the remaining “villages” themselves museums of cultural appropriation and the use of Native American imagery in popular culture.

Students with Dr. Algeo (center) at Wigwam Village No. 2, Cave City, KY

Following supper, we returned for a second session with Dr. Algeo by the “campfire” (which is lit in a sinkhole and has been a feature of motel stays since 1937). Her engaging talk considered the significance of Mammoth Cave to successive populations, its development as a symbol of national natural/sacred heritage, and how the use of local caves for T.B. treatment and for tourism in the nineteenth century related to contemporary views of race and class. Literally reflecting on ‘perspective taking in context,’ all students had questions and the conversation lasted for nearly 3 hours. Breakfast the next morning at an Amish bakery involved, for some students, dialogue with staff of a similar age about their views of Mammoth Cave and how daily interactions with tourists articulated with their community life, while others focused intently on the sweet fare. By 9AM we were on our way for a Lantern Tour of Great Onyx Cave to experience “sacred darkness” and the sensory load of “being” deep within a cave. The visit was the culmination of class lectures, student research, and discussions about sacred caves around the world and also bridged to class coverage of current invasive tourism at Sacred Natural Sites globally.  

by Mark Hopwood

There is no silver bullet for ChatGPT. Nothing that we can do will make our courses and assignments completely AI-proof, and no AI detection program is completely reliable at identifying AI-generated text. According to James Lang (the author of Small Teaching, Cheating Lessons, and many other influential books on pedagogy), these facts should not cause us to despair. Even if we cannot completely rule out the possibility that students will cheat, there are plenty of relatively simple things we can do to make it much less likely that they will be tempted to do so. Furthermore, the fact that AI can be used to complete a particular assignment does not mean that there is no value in teaching students to do it for themselves. If anything, the advent of large language models only serves to underscore the importance of some basic principles that have always been true of college teaching: 

(i) students will do better—and be less tempted to cheat—when we are clear and transparent about what we are asking them to do and why it is important for them to do it, and 

(ii) giving students the opportunity to engage in metacognitive reflection on their own learning process not only helps them to retain what they have learned, but also to see the value of doing the work for themselves.

These insights were some of my main takeaways from a webinar delivered by James Lang on “Academic honesty in the age of AI” earlier this semester. (A full recording and slides are available here.) I found the webinar very helpful, both for the measured and realistic discussion of the potential impact of AI on college teaching and also for the concrete suggestions offered for how we can reduce the temptation to cheat while also improving students’ understanding of their own learning. One particularly interesting takeaway was Lang’s discussion of a recent study of a range of “low-effort cheating-reduction methods”, which showed that strategies as simple as reminding students of the AI policy midway through the semester (rather than mentioning it only once at the beginning) and writing transparent assignments that explain why students are being asked to do what they are being asked to do made a significant difference to the incidence of cheating in intro-level computer science classes. Of course, none of these strategies make it impossible for students to cheat, but as Lang noted, students have always cheated, even before the emergence of ChatGPT. Protecting academic integrity is critical, but it is also important for us to remember that our main job is the same as it has always been: helping students to learn. What was perhaps most helpful about the webinar was the way in which Lang demonstrated that promoting student learning and protecting academic integrity need not be in tension with each other. In fact, the very same strategies that help us to achieve the first aim will also help us to achieve the second. Variety in teaching methods, transparency in assignments, and opportunities for metacognitive reflection have always been basic principles of good pedagogical practice, but in the age of AI they have become even more essential. 

I highly recommend watching the recording of the webinar and trying out some of the specific strategies Lang recommends in your own classes. If you have tried out any ideas to help your students learn while also reducing the temptation to cheat, let us know: we’d love to feature them in a CfT blog post or session!

Maimonides grading student papers on a laptop (Image created using ChatGPT / DALL-E)

by Ian Jensen

What Can LLMs Do for Me?

If you’ve already read the first part of this guide, you might quite reasonably be asking “So what [legitimate] uses do these LLMs have?” It’s a good question and I’m not entirely sure there’s a clear answer at this time, except that LLMs are useful insofar as you need to produce fairly large amounts of grammatically correct text quickly. What the AI people keep telling us is that LLMs are able to e.g. A) give information clearly, B) help us generate ideas, C) help us revise our writing, D) summarize ideas and papers/articles, E) write emails, and so forth.

As we’ve already seen, I’m skeptical about A and B. Since the LLM has no sense of what it is generating it’s all too easy to get incorrect information from it. The same issue applies to helping us with ideas. Since the LLM has no ideas, since it doesn’t connect its tokenized version of language to anything other than its combinatorial options, whatever “ideas” it generates are only perceived by the reader and that reader might better spend her time looking at the actual material rather than querying chatbots. 

Helping us revise may in fact prove useful, although at this time it’s difficult to retain or produce a writerly “voice” after you’ve run your text through the algorithm. To the extent that for work emails voice is immaterial, there is a use case here. At the same time, by the time you get what you want from the LLM (see below) you probably could’ve easily written the email yourself.

So we’re left with D, summary. I won’t get into details here, except to say that while any given summary of a text already in the AI’s dataset or one you’ve uploaded may turn out to be accurate, there is no guarantee. Because of the ways in which LLMs do summary, they can either try to extract and reword key ideas/phrases/sentences from the original text (your document or a Wikipedia entry) or they can process the entire document, compare it to the pattern and probability datasets the LLM has trained on and then attempt to condense and state the main points. 

In either case, it all depends on how well the LLM is able to process the text it is provided, which means two things. First, the more straightforward that text is, formally speaking, the easier it will be for the LLM to accurately summarize it. Second, the LLM will do a better job of summary when it has been trained on many similar documents. Since much of the text that we rely on in the humanities is far from straightforward–one shudders to imagine ChatGPT processing English translations of German cultural theory or highly figurative, allusive writing like that of Homi K. Bhabha and Hortense Spillers–the summaries we’re going to get won’t be very satisfying or even accurate. 
It’s worth noting here that paywalled databases of academic work–Project Muse, to take an example–have not been crawled, scraped, or linked to publicly available chatbots. Those databases will eventually be scraped and linked to LLMs. I see it going one of two ways. In the first scenario, tech companies pay large amounts of money to scrape academic databases for, say, AI-powered Google Scholar. Second, the databases license or buy LLM AI programs from the tech giants and then bake them into their databases. Either way, the amount of money to be made by middle-persons on our unpaid labor truly boggles the mind.

Predictions

Prognostication is a notoriously fickle business, but it is certainly interesting to think about where all of this is going. I do think that what we might call the routine writing of business and education –emails, memos, summaries–will largely be done by AI in the coming years. It is important to note that at this point, generative AI doesn’t threaten the humanities in terms of its usual benefits insight, research, argumentation, critical thought, or topicality in any way–other than incentivizing widespread academic dishonesty. 

Unfortunately the humanistic skills are often not valued in our present economy or society, so I don’t have much hope for case that the humanities ought to be supported because they are bastions of “good” writing and thought. Students who are rightly focused on future employment (as many of them are) are generally going to struggle seeing the relevance of what the humanities and liberal arts offer them in their career searches. The truth of the matter is simply that most people don’t end up needing calculus or Latin directly, and I worry that even the very basics of the humanities writ large (rhetoric, rigor, argumentation) will be seen in a similar way. (NB: Clearly utility and value are not synonyms and I’m not suggesting that learning things without obvious “practical” value is pointless. But given the current economic situation, both in terms of cost of college and finding gainful employment after college, it’s a harder sell.) We can argue that the humanities “good for you” until we’re blue in the face, but outside pressures trump idealism most of the time—as they probably should:  “The confusion over the purpose of the humanities has nothing to do with their relevance. The humanities are no more or less relevant now than they ever were. It is not the humanities that we have lost faith in, but the economic, political, and social order that they have been made to serve. Perhaps we demand a case for the humanities only because we cannot fathom having to make a case for anything else.”

In a more hopeful vein, it’s possible that publicly available untraceable  LLM chatbots turn out to be to be a flash in the pan. One is reminded of the earliest years of sampling in music. The common view is that sample-heavy records like De La Soul’s 3 Feet High & Rising and the Beastie Boys/Dust Brothers’ Paul’s Boutique are essentially impossible to produce now (as legal, mass market works) for copyright reasons. The business of web crawling and scraping vs. copyright is a major one and it could all but kill off companies and products like OpenAI and ChatGPT. Given the immense investments every big tech company except Apple seem to be making in AI, this is probably a very unlikely outcome.

Postscript – What Can Teachers Do?

There have been lots of hot takes on this topic over the past year at the Chronicle and its ilk, as well as on Reddit, blogs, Twitter, and elsewhere. I’m not sure much of it is deeply useful–I’m skeptical of ideas like “flipping the classroom” and also of making my classes into what are essentially AI use trainings. I’m also not going to make students handwrite 4+ page essays in class, although I won’t be offering take-home exams either. There’s the idea that showing students how bad chatbot responses can be will be useful, but to me spending any more than half a class on that exercise seems excessive.

There are, as mentioned, free-ish AI detectors out there but they are said to be highly unreliable. GPT-Zero has gone so far as to create a Chrome browser extension that will record student work on Google Docs and send you a video of the composition of an essay. I hope never to have to mandate its use, but we’ll see how the semester goes.

Good luck out there and godspeed.

• Find a list of administrative & course policies at a wide variety of institutions here
• Microsoft’s AI (GPT4) Page 
• Microsoft launches GPT4-powered  “Copilot” app for smartphones

“Maimonides entering a query into ChatGPT on a laptop” (Image created using ChatGPT/DALL-E)

by Ian Jensen, Department of English

Introduction

The title of this guide alludes to Maimonides’ influential development of apophatic/negative Jewish theology in the 11th century. While for Maimonides God can only be described in negative terms, here I’m concerned to develop a working account of the kinds of “AI” we teachers of writing are likely to face from now on. As I hope to demonstrate below, however, what large language models  (LLMs) aren’t is equally as important as what they are. 

I’ll begin by explaining how LLMs work without a lot of jargon or computer science. I’ll then discuss what they are (text generating machines), what they aren’t (intelligent in any meaningful way), and what all this means for us. 

Finally, two things to keep in mind. First,  I am no expert here, so this is one person’s pieced together account. As a result its accuracy may not be 100%. Second, this will be an opinionated guide. Caveat lector!

What are Large Language Models?

It’s important to remember that generative artificial intelligence (AI) is a genus, and that large language models (LLMs) are a species of AI. Here  I’ll focus on the biggest issue for writing teachers, large language models. 

Elastic notes that “ A large language model (LLM) is a deep learning algorithm that can perform a variety of natural language processing (NLP) tasks. Large language models use transformer models and are trained using massive datasets — hence, large. This enables them to recognize, translate, predict, or generate text or other content.” It is important to understand that LLMs are essentially algorithms, not so different from the predictive text typing we find on our phone messaging apps or the music predictions produced by Spotify etc. 

Ok, so let’s define algorithms. The International Institute in Geneva says that  “[A]n algorithm is a set of instructions, used to solve problems or perform tasks, based on the understanding of available alternatives.” Algorithms process data and apply rule-based frameworks (e.g. English grammar and syntax) to that data. From these, the algorithm is able to “predict” what is likely to follow a given relevant input. So e.g. when you “like” a post on Instagram, your interest is fed into the algorithm which will then populate your feed with posts it deems to be similar. The same goes for advertisements we see in Google, Facebook, and so forth.

Generative AI LLMs take these “predictive” algorithms and trains them on more or less the entire free internet while  exponentially expanding the parameters that the algorithms are able to take into account. Rather than “predicting” a given word based only on the word before it, which is what many of our messaging apps seem to do, LLMs are able to widen the parameters of how much data is taken into account. So instead of using a single word, an LLM might be able to use a full sentence, a paragraph, or even an entire essay or book “in mind” as it “predicts” what is likely to come next. I’m using the example of text here but AI algorithms can be trained on images, sounds, biological data, and so forth. Google’s AI division has already developed a multimodal generative AI they are calling, unsettlingly, “Gemini.” 

Now back to deep learning, which simply means that rather than having only one data processing “layer” there are multiple layers that increasingly refine the data processing that is used as the basis of “predictive” generation. These layers are called neural networks. The point is that the “predictive” algorithms of large language models have processed inconceivably large amounts of data (i.e. text) in a very sophisticated way called “natural language processing.” From that data and the parameters both embedded into them by humans and self-produced from data analysis, the LLM can then produce astonishingly coherent text. 

OpenAI’s GPT-4, the one that powers paid ChatGPT and Bing Chat Microsoft Copilot, famously used a method called “crawl-and-scrape” to train. Essentially, OpenAI built script-running “bots” that index and scrape data from publically available websites (and other resources). If this sounds ethically questionable, that’s because it is. Indeed copyright lawsuits have been and are being filed, some by big name authors & publishers, as well as the New York Times: their complaint is well-worth a read. In addition, publicly accessible large repositories of human-produced text like Wikipedia and Reddit are beginning to close ranks against LLM-training crawl-and-scrape bots, in one case causing Reddit to alienate a number of contributors (redditors) and communities over the summer. 

NB: Anything you have written that has been posted to a publicly accessible website has likely been used to train at least one LLM. This includes, but is not limited to, open access papers, your CV, syllabi, profiles at Academia.edu/ResearchGate/Google Scholar, personal blogs, Substacks, lectures posted on YouTube, comments you’ve posted, and so much more. Welcome to the brave new world.

Are Large Language Models Actually Intelligent?

The short answer is no, they are not. The reason for this leads me to the longer answer, which is absolutely crucial for both teachers and students to understand: Large language models, and indeed all generative AI so far available to the public, does not in any sense understand what it processes or produces. 

In fact, in LLMs the text generated is produced through something called “tokenization.” In tokenization, the letters and words produced (and processed) by  the LLM are assigned non-semantic data values, with non-semantic here meaning having reference to meaning. From there, the LLM produces the next letter or word token, which is based on the prior tokens in the model. 

What this means for teachers is that generative AI produces pure text devoid of understanding, intent, or semantic content. This empty text based on probability is nothing other than a product of tokens assigned lexical skins, as it were, being strung together. As a result, the text produced is not in any usual sense a message nor does it have content. Any meaning that the text appears to have comes entirely from the reader. There is no ghost in the machine.

When initially made available, ChatGPT was not connected to the internet and relied on training data from its initial “crawl and scrape” of the internet. Increasingly, LLM chatbots are being connected to the internet—ChatGPT’s paid service, Google’s Bard (i.e. Gemini), and Microsoft’s CoPilot (formerly Bing Chat & powered by OpenAI’s GPT-4) are all able to integrate real time data from the internet. This is bad for instructors because it mitigates some of the errors seen in early LLM output, which in turn makes IDing LLM text more difficult.

All of this is important to remember when pundits and AI CEOs tell us about all of the creative work LLMs will help us do, and helps us understand why LLMs “hallucinate” false references and factual errors. In the first case, and keeping in mind that generative AI has been and will be useful in many capacities, it is simply the case that in today’s LLM-generated text is produced out of thin air as it were–a thousand neural network monkeys on typewriters so to speak. Certainly this doesn’t foreclose the possibility of inspiration striking a reader of LLM generated text, and perhaps more generously we could think of LLM text as something closer to Burroughs’ cut-up experiments in such a use case, but that inspiration is not the product of any kind of linguistic function Jakobson would recognize–there is no sender and there is no message; there is no line of communication to open, etc. 

In the second case, LLM hallucinations occur precisely because of predictive tokenization and the absence of knowledge or understanding. So because Stephen Greenblatt and Julia Lupton have written a great deal about Shakespeare, if we ask an LLM to add citations to our essay about The Tempest it is completely possible that it will do so by simply generating a tokenized name like “Greenblatt” or “Lupton” and a page number that fits the pattern it trained on: so never a 1, 2, or 3; never exceeding 600; falling between say 200-500, etc. It may also very well generate the title of a book or paper that fits its recognized pattern for something like “titles associated with Shakespeare and including an author reference to Lupton.” You could well get something like [“Affects and Affordances in Hamlet” (2022)] as a reference, despite no such article existing. 

For more on what LLMs can do and how we might think about them in the college classroom, look out for part 2 of this guide on The CraFT next week!

Kate Cammack & Kristen Cecala

Was there a specific point in time where you felt like you might be a “science person”?

What experiences – positive or not – contributed to your sense of confidence and ability within STEM?

As educators, we have a precious opportunity to cultivate & encourage students’ interests, sense of self-efficacy (belief that they are capable and can succeed), and belonging in our learning spaces.

Within STEM disciplines, expanding students’ sense of belonging helps to broaden the diversity of students who explore, pursue, persist in, and complete STEM majors/minors. It also enhances students’ “buy-in” to introductory (often G5-linked) STEM courses, which builds a more scientifically literate citizenry regardless of their eventual major/minor. Both require students seeing themselves as capable and able to succeed in STEM.

But there’s a lot to learn, and little of what we need to know was taught to us – or modeled – in our STEM training.

This past fall, eight Sewanee faculty, representing seven of our STEM departments/programs, joined our first faculty learning community (FLC) on DEIBJ in STEM. Our FLC goals were to (a) identify, share & discuss evidence-based approaches that enhance DEIBJ in STEM, (b) cultivate comfort and confidence to implement these approaches, and (c) develop changes/deliverables to implement in our classrooms.

We used the Inclusive STEM Teaching Project (a 6-week online edX course) as a framework for in-person discussions and workshops. We all completed that week’s edX materials (see below), then met in the CfT for an in-person debrief and discussion. Our in-person meetings allowed us to experience productive discomfort and personal growth throughout our edX experience, and to contextualize this content in ways specific to our Sewanee community. This work continued for 6 weeks, from Sept – early Oct.

While this edX course is offered regularly throughout the year, our FLC structure enabled us to engage off-cycle, at a time that works better for Sewanee faculty. It also allowed us to meet in person to discuss the Sewanee-specific aspects of our learning and discuss the specific opportunities facing our community with respect to DEIBJ in STEM.

This FLC will be offered again in early Fall 2025, and hopefully regularly after that – please consider joining us. It’s been a valuable way to engage with colleagues across disciplines, find common goals & synergies, strengthen and expand our on-campus network of DEIBJ-engaged STEM faculty, and contribute to transformative institutional change. You are needed!

Stay tuned for more info to come, including a Brightspace page dedicated to teaching & learning resources for DEIBJ in STEM. We’re also starting to build capacity with the support of a 2024 McCrickard Grant and welcome your contributions, thinking, and help.