When text becomes data: how AI is changing research, teaching and libraries

AI is reshaping research and how data skills are taught. The potential is clear, but so are the questions. What sources can be trusted? How do researchers move from experimentation to rigorous analysis? What role should libraries play as these tools become part of everyday academic work, and how can they support this shift sustainably?

The Pulse of the Library report from Clarivate shows many libraries are already navigating this shift. More than two-thirds are exploring or implementing AI, and over half say it will require significant upskilling. Ethical use and AI literacy remain top priorities.

The webinar From News to Data: AI and Text Analysis in Research, Teaching and the Library offered a view of how this shift is playing out across higher education. The discussion moved past AI hype to a more practical question: what does it take to use these tools well in research, teaching and library work?

Speakers included Michele Hayslett, Librarian for Numeric Data Services and Data Management at the University of North Carolina at Chapel Hill, Johan Cassel Pegelow, Assistant Professor of Finance at Vanderbilt University, Yifei Lu, Assistant Professor of Accountancy at the University of Illinois Urbana-Champaign, and Tarika Sankar, Digital Humanities Librarian and Lecturer in Humanities at Brown University. Together, they showed how AI is reshaping academic practice.

Looking Critically At AI Output

AI is expanding what researchers can do while raising the bar for judgment. It can speed up discovery and analysis across enormous collections of text, but scale is only useful if the results can be trusted. For libraries, that means the work is no longer just about providing access to tools. It also includes helping students and researchers question what those tools produce, while keeping training current and supporting AI use in ways that are sustainable for library staff.

As Hayslett put it: “We’ve always been in the business of teaching people about critical evaluation, but it’s even more so with these tools.” AI may be new, but the need for source evaluation, context and informed use is not.

Creating A More Structured Workflow

There’s also the growing importance of turning unstructured text into usable data. Pegelow used newspaper content to show why this matters. In his research on private markets, he demonstrated how AI can extract structured insights from unstructured news to analyze events like venture capital funding rounds. Using examples such as SpaceX and OpenAI, he showed how newspaper coverage can be transformed into data that help researchers track investment activity and better understand how these markets operate.Tools such as the TDM Studio text and data mining tool can help researchers build datasets, apply prompts and review results within a more structured workflow. But the larger point is methodological. AI can assist with extraction and organization, yet careful design and validation still determine whether analysis holds up.

That is what makes this kind of work credible: not just using AI, but building a process for testing, refining and checking what it returns.

Redesigning Courses For What Students Need To Learn Now

That shift is also showing up in the classroom. Lu described how AI changed the design of his graduate course in data analytics for accounting students and what students now need to learn: how to ask better questions, work responsibly with data and make sense of what the analysis shows.

He shifted the focus of his course from coding to the full analytical process: framing the question, finding data, designing prompts, validating outputs and interpreting results. Using news datasets and TDM Studio, students explored topics such as college athlete compensation and Bitcoin, developing their own questions instead of following a set sequence. The result was a more practical approach to data literacy tied to inquiry, interpretation and decision-making.

Lu said students responded strongly because the approach felt closer to the AI-enabled work they expect beyond the classroom.

Improving AI Literacy By Understanding How The Tools Work

If AI is becoming part of academic work, then AI literacy has to go beyond learning how to prompt. Sankar made the case for more critical engagement by combining hands-on use with deeper reflection on how these systems work and where they fail.

That means treating AI as something to interrogate, not just adopt. At Brown, Sankar is leading workshops that ask students to build datasets, examine what is missing and reflect on the ethical and analytical limits of what they have created. The goal is not just fluency with tools but the ability to see how data choices shape outcomes. “Some of the most productive engagement is when you are bringing the theory and the practice together,” she said.

A Changing Research Landscape

AI is making it easier to analyze text at scale and turn it into structured data, but it is also making rigor, transparency and critical judgment more important.

For researchers, that means clearer methods. For students, stronger analytical judgment. For libraries, an expanded role in helping their communities use AI with confidence and care. The tools may be changing quickly, but the questions of evidence, interpretation and trust remain central.

Watch the full discussion in the on-demand webinar.

If you want to explore these workflows further, learn more about TDM Studio.