Over the last few years, professors have asked us how to approach research assignments in light of recent advancements with AI. While the tools aren’t perfect, research assistants like Scite, Consensus, and Elicit produce impressive-looking outputs that synthesize evidence from across multiple sources. The introduction of these AI research assistants raises many questions – one being whether students should even use them, and if so, for which uses might they be okay and for which uses might they not. These tools have also prompted us to revisit a more fundamental question: What do we want students to learn from researching scholarly literature in the first place?

If you’ve also been wondering about research projects in the AI era, Kathy and Amanda hosted a First Five: ZSR Early Career Faculty Workshop on this topic. What we’ve included below is a snapshot of our five key recommendations, which range from looking at the “big picture” of why we assign research assignments to practical strategies you can implement in assignment design.

“Big Picture”

1. Discuss the Purpose of Research

The first thing we recommend might sound overly simple, but it involves helping students understand why we research scholarly literature in the first place. As faculty members, it may seem obvious to us, but to students early in their research journeys, it may not be. Revisiting the purpose of scholarly research may also mean reflecting on assumptions long held, and pausing to consider the “why” behind things that are routine. Why do we need to know and understand what’s already been studied or discussed? What would we gain and lose from having AI summarize and synthesize previous research for us?

We recommend asking students these questions explicitly and seeing what they say. Often, their responses give us a starting point for discussing the value of research and allow us to fill in gaps in their understanding. We can explain that while AI can change how we encounter academic scholarship, it has yet to change the fact that knowledge is built collectively, through an ongoing conversation among scholars that builds on the knowledge of others. Different scholars often contribute complementary pieces to a puzzle, and part of the purpose of student learning during a research project is figuring out how those pieces fit together, where they don’t, where there are gaps in the literature, and where they might begin to construct and articulate their own perspectives and interpretations. When an AI tool summarizes and synthesizes the literature for us, it makes those judgments on our behalf. Yet those judgments are the very skills we hope students develop through doing research.

2. Help Students Consider Where AI Might Help and Harm

On the heels of a conversation about the purpose of scholarly research, it can be helpful to have a conversation with students about where AI might help or harm our learning within the research process. One way we’ve done this is by breaking the research process down into its various steps and considering the role of AI in each step individually. Different models exist for describing the steps in the research process, but our model generally begins with identifying an area of inquiry and ends with reflecting on the research process as a whole.

For this discussion, we invite students to form small groups and consider the following uses of AI and give them an “A” to an “F” on the grading scale. Students are also asked to provide justifications for their “grades,” and we remind them that there are no “right” or “wrong” answers to the questions in hopes that they are more honest about their opinions. Here are the uses we ask them to grade:

• Posing Research Questions: Using AI to help us brainstorm topics for our research projects
• Finding Information: Using AI to help us find high quality information
• Evaluating Sources: Using AI to help us evaluate whether a source is biased, reliable, or suitable for our research topic
• Reading and Analyzing Evidence: Using AI to help summarize sources we’d like to include in our research
• Synthesizing Sources: Using AI to help us synthesize information across multiple sources
• Giving Credit: Using AI to help us cite our sources
• Reflecting: Using AI to reflect on your research process to determine what went well, what didn’t go well, what to research in the future, etc.

Students (and professors) often have varying opinions, so it’s helpful to prepare for a variety of scenarios. If it’s feasible, inviting another professor or librarian along for the discussion can add a rich layer to the discussion. When two instructors grade the same use of AI differently and explain why, students get to see the kind of scholarly conversation we described above play out in real time. For example, students are often surprised to hear that we don’t always give AI an “F” for finding scholarly information (after all, AI is maybe our greatest hope yet for an end to boolean searching 😂) but many professors maintain that search friction can be a form of productive struggle that AI bypasses. Either way, these questions allow for open discussion of the pros and cons of using AI for research.

Practical Strategies

In addition to making the value of research explicit for your students in your particular course, there are also some practical strategies that you can implement to help students spend time developing and practicing the research skills you want them to cultivate. Depending on the goals of the assignment or the course, you can determine which strategies best meet your needs.

3. Scaffold Assignments

Scaffolding assignments involves breaking an assignment down into smaller components and creating check-in points to make sure that students are on the right track, rather than having them turn in one big assignment at the end of the semester. Scaffolding has been recommended for a long time for complex research projects, and it remains a useful tool. One of the key benefits of scaffolding as it relates to research assignments in the era of AI is that it helps make the research process more transparent, both to the professor and the student. It allows students to articulate their thought process and any areas of struggle or friction and allows you to provide direct feedback and guidance at each stage. While these strategies are particularly useful for students who are unfamiliar with academic research, they are equally helpful for upper-level students who are learning how to dig deep into the literature of their discipline.

Some ways to add scaffolding to a research assignment include:

• Research Log: Students use this to track their search process and strategies. You can also ask them to identify any challenges they come across in their search process or other information that is specific to your assignment. (Example
• Annotations: Students turn in an annotated bibliography of the sources they intend to use for their research project. In addition to a written summary of the source, ask students to include additional required information, such as a specific quote (with page numbers) that relates to their research question.
• Synthesis Matrix: Students discover themes through creating a synthesis matrix. While there are AI tools that can do something similar, you can again require elements that are specific to your assignment. (Example
• Research Reflections: Students periodically reflect on their research process – what went well, what didn’t, what they learned, how their thinking changed, etc. These do not have to be long, and you can even do these in class, on paper.

In contrast to the “black box” nature of AI tools, through these kinds of assignments, students have to “show their work” and demonstrate their research process – how they found a source, why they chose it, how it relates to their research question, and provide a personal reflection on the experience. And, it allows you to see how their research progresses over the course of the semester.

And while most of us aren’t keen to spend time policing students’ use of AI, these scaffolded assignments can also provide an opportunity to help students get back on the right track and revise their research process if it becomes clear that they are relying on AI too heavily. One thing to be on the lookout for is what some scholars are beginning to call the “missing middle,” or literature searches that are characterized by the citation of several older, foundational papers on the topic and then several recent papers, which are often from open access journals or pre-prints from websites like arXiv. This “missing middle” happens because AI searches using popular generative AI platforms like Claude and Gemini tend to draw on what’s openly available online. These literature searches rarely contain paywalled academic sources and have limited sources from the mid-90s to the early 2020s. When we see this pop up in scaffolded assignments, we use it as an opportunity for learning – to talk about what’s missing and how additional research (using library resources) can help fill in the gaps.

4. Focus on Deep Reading

Our next recommendation focuses on teaching the skill of deep reading. Elizabeth Ellis and Amanda recently described how students in our credit bearing program have reported one of their top information literacy struggles as being, “attention span,” “reading comprehension,” “reading long pieces of information,” “patience with long readings,” and “focus in the age of distraction.” This tracks with recent reports on the declining rate of reading and reading comprehension among Americans, and importantly for us, among college students. This presents a challenge for students reading scholarly literature.

One of the ways that you might focus on deep reading is by modeling how you read scholarly literature. Explain which sections you look at first and why. Outline the questions that you are asking yourself as you read and specific elements that you are looking for. Be clear that these are skills you’ve developed as an expert and make explicit what’s expected of them as emerging scholars. It’s also important to remember that students do bring strategies and skills with them from high school and other courses. Ask them to articulate those strategies and how they can build on them to read this new (to many of them) genre of writing or apply them to scholarly literature in a different discipline. If you prefer to build off what other professors have done, we recommend Keshav’s Three Pass Method for reading a paper as a good starting point.

5. Have Students Annotate, Highlight, and Turn In Their Printed Sources

Especially when working with students who are not as familiar with the genre of academic writing, primary sources, or whatever other sources you use in your discipline, it’s beneficial to give students opportunities to practice analyzing them. While hand annotating and highlighting may seem basic, most students are not taught how to use them effectively to support deep engagement with a source.

• Ask students to print out and hand-annotate a source. Maybe start with a course reading, then build on this in future assignments. (This can also be adapted to a digital format to support student accommodations with tools like Zotero.)
• Create an annotation log/legend (individual or group activity). This can reflect specific elements required in the assignment or elements that the group decides are important:
  • Identify main thesis or hypothesis
  • Identify articles from the literature review they might want to follow up on
  • Identify quotes from the discussion section (or deep in the article if there’s no discussion section) that are relevant to their research question
  • Identify points of confusion – ask questions in the margins
  • Identify connections to other ideas or readings through marginal annotations
• As an alternative (or complement) to annotation, ask students to create a concept map of the article that breaks down key arguments and evidence.

In another effort to make their research process and evidence visible, when students turn in their annotations or synthesis matrix or even their final project:

• Ask them to include copies of the sources they used (hard copy or PDF) and to highlight in the sources where their quotes/paraphrases came from.
• Require page numbers for quotations/paraphrases and URLs even if the citation style doesn’t ask for them.

When you make this a regular practice in your course and stress the purpose of building these skills, it becomes less about policing and more about helping students to develop their capacity to understand and synthesize scholarly literature.

Conclusion

We recognize that faculty approach AI from a range of perspectives. If you see something here that resonates with you, or you want to discuss strategies for your specific context, we would love to hear from you! Reach out to your library liaison or contact one of us with questions.