AI can now do things that would have required a team of technical specialists a few years ago, or that simply were not possible at all. Humanists can explore patterns across thousands of pages of archival material without writing a line of code. A literature seminar can analyze an entire genre. An oral history project can make decades of recordings searchable in an afternoon. A student can build an interactive archive as a course project that would have taken a development team two years ago.
That’s a genuine shift—not just in what’s accessible, but in the kinds of questions humanists can ask in the first place.
That reach comes with real costs, which we’ll get to. But the starting point is possibility: AI has opened research territory that was previously out of bounds, and we want to help you explore it.
Human-centered AI at Amaranth means the person doing the research stays responsible for the question, the evidence, the interpretation, and the audience. AI can help search, sort, describe, compare, and draft. It cannot decide what matters.
That is why we care about provenance and documentation. A useful AI workflow should make its steps visible: what sources were included, how they were prepared, what the model was asked to do, where the output came from, how it was checked, and what claims survived human review. If no one can explain how an AI-assisted interpretation was produced, it is not ready to become scholarship.
Some projects also need researcher-controlled or institutionally accountable AI. Community oral histories, Indigenous collections, unpublished archives, and sensitive cultural materials should not be treated as generic input for whatever tool is convenient. Human-centered AI asks who has a stake in the material, what permissions matter, and how the process can remain answerable to the people whose histories are involved.
Exploring patterns across texts. You have hundreds of newspaper articles, letters, government documents, or literary works. AI can help you identify recurring themes, trace how language shifts over time, compare rhetorical strategies, or surface connections you might not have found through close reading alone. You still decide what the patterns mean and which ones matter—but you can now see across a collection at a scale that changes what’s possible to ask.
Transcribing and searching oral histories. You have hours—maybe dozens of hours—of recorded interviews. AI can transcribe them quickly and let you search across the full collection: every mention of a place, a practice, a name, an emotion. The transcripts need human review (AI stumbles on names, accents, and specialized terms), but starting from a draft rather than silence saves enormous time and makes large collections usable in ways they weren’t before. For community oral history projects, materials that were difficult to use can become navigable.
Analyzing images and visual collections. You’re working with photographs, artworks, maps, or material objects. AI can help you identify visual patterns, compare compositions, tag and categorize at scale, or generate descriptions that make visual collections searchable by text. For digital exhibits, AI can also help you create or manipulate images to support your argument or narrative.
Research assistance and literature mapping. Starting a new project or entering an unfamiliar field? AI can help you map the intellectual landscape—identifying key debates, summarizing major positions, suggesting search terms, and pointing toward sources you might not have found on your own. Think of it as a well-read but unreliable research assistant: useful for orientation, always in need of verification.
Structuring messy data. Archival material often arrives in inconsistent formats—variant spellings, mixed date formats, incomplete records. AI can help normalize and structure that data so you can actually analyze it, map it, or visualize it. This is often the unglamorous work that makes a digital project possible.
Thinking and writing tools. You have a rough argument and want to pressure-test it. AI can help you identify gaps in your reasoning, suggest counterarguments, reframe your claims for different audiences, or help you restructure a draft. This isn’t about AI writing for you—it’s about using AI to think more rigorously about what you’re trying to say.
Student projects that would have been impossible. With some guidance, students can use AI to build things that previously required significant coding skills: searchable archives, interactive timelines, annotated maps, text analysis projects, even simple web applications that present research to real audiences. The technical ceiling for humanities projects is genuinely higher than it was two years ago.
Community and interdisciplinary projects. Some of the most interesting work happens at the edges of disciplines and between the university and the communities it exists alongside. AI has particular value here: oral history collections in community hands can become searchable archives; materials that were sitting in folders can be analyzed, published, and shared. We actively seek out collaborations that cross departmental lines and connect scholarly work to broader audiences and partners.
The workflow itself can be a research output. Not just the article, exhibit, or website that comes at the end, but the documented record of how the work was done: what was tried, what the model produced, what failed, what needed human correction, and how someone else could reproduce or adapt the method.
This is the practical layer the field needs more of. Humanists do not only need claims about AI. We need examples of careful practice: prompts, source preparation, model limits, verification steps, and plain accounts of where the work got strange. That documentation is part of the scholarship because it lets others inspect the method instead of trusting the result.
We want to be part of the research conversation early, not brought in only for technical cleanup at the end. Come with a half-formed question, a dataset you’ve never figured out how to approach, or a sense that your materials could do more than they currently do. We’ll think with you about what’s possible, design a workflow that fits your project, iterate together, and get the work into a form that can be shared and preserved.
We work alongside you and learn what these tools can do in the context of your particular materials, questions, and audience. Context changes everything.
We start where you are. Many of our collaborations begin with something technical and low-stakes: helping build or update a website, formatting documents, troubleshooting a project structure. Not because that’s our real interest, but because technical work with clear feedback loops—you can see whether the font changed, whether the site builds correctly—is a genuinely good way to get comfortable working with AI before the intellectual stakes are higher. Once that comfort is established, the step into research workflows feels much smaller. We’ve found this progression matters.
Interdisciplinary and community collaborations. The questions humanists ask rarely stay within departmental lines, and neither do we. If you’re working with colleagues from another field, or with a community partner whose history and materials deserve scholarly attention, we want to hear about it early. The earlier we can be part of the conversation, the stronger the project.
The goal of working with AI isn’t to hand off intellectual work. It’s to expand what you can do and ask—and to build the judgment to know when AI is helping and when it’s misleading you.
Faculty who work through AI-assisted research projects develop the kind of critical fluency that shapes how they design courses and advise students. Students who build something real—a text analysis project, a searchable oral history collection, an interactive archive—leave with skills and confidence that transfer well beyond any single assignment. They understand what it means to direct a technical workflow, evaluate its outputs, and take responsibility for the argument.
Humanists are particularly well-positioned for this moment. The interpretive habits built through close reading, archival research, and critical analysis—evaluating sources, questioning frames, recognizing whose voices are missing—are exactly what separates meaningful AI use from mechanical AI use. We help you see that you’re already more prepared than you might think.
It’s worth being direct about this, because the hype around AI tends to skip these parts.
AI doesn’t understand your material. It processes patterns in language and data, but it has no sense of historical context, cultural significance, or why something matters. It will give you a confident answer that’s completely wrong. It will summarize a complex debate into a tidy paragraph that loses everything important. It doesn’t know what it doesn’t know.
AI fabricates sources. This is not a rare glitch—it’s a fundamental feature of how these tools work. If you ask AI for citations, it will sometimes invent books and articles that don’t exist, complete with plausible authors and publication details. Always verify.
AI reflects the biases in its training data. The material these models learned from is disproportionately English-language, Western, recent, and digitized. Marginalized perspectives, non-Western traditions, oral cultures, and anything that wasn’t well-represented online will be underrepresented or distorted in AI outputs. For humanities work—where these gaps and biases are often exactly what you’re studying—this matters a lot.
AI flattens nuance. It’s very good at producing clear, confident, well-structured text. It’s bad at ambiguity, contradiction, irony, and the kind of complexity that makes humanistic inquiry interesting. If you use AI to summarize, you’ll get a summary—but you may lose what made the original worth reading.
Use AI the way you’d use any powerful but imperfect tool: with clear eyes, critical judgment, and a humanist’s sense of what matters. Recognizing these limitations clearly—and being able to explain them to students and colleagues—is itself a humanistic skill. Faculty and students who develop this critical fluency can help their institutions and communities navigate AI more thoughtfully. These conversations happen constantly at Amaranth.
You don’t need a technical background. You don’t need a fully formed project. You don’t even need to know which AI system to use. If you have a question, “could AI help me with this?”, that’s enough.
Drop by studio hours. Tuesdays & Thursdays 9:30–11:00 and 12:30–2:00, Wednesdays 10:00–12:00. Bring your laptop, or use ours. Bring a question, a dataset, a hunch. We’ll explore it together.
Book a consultation. If you’d rather talk through your project before diving in, book a consultation and we can figure out whether AI fits, how it fits, and what the project might become.
Email us. amaranth@unm.edu. Even a one-line question is a great place to start.
Read about how we investigate and document our own AI practice. In our AI Sketchbook, we share what we’re learning semester by semester—what works, what doesn’t, where the edges are.