In 2022, I shared the idea of critical thinking and ignoring as a part of a CETL Teaching Tip, built using Mike Caulfield’s SIFT Method for evaluating sources (stop, investigate the source, find other coverage, and trace ideas back). Since then, the information landscape has changed dramatically: Generative AI tools are everywhere, even when our students don’t see them mediating the information sources that our students encounter. What can we do now to help students engage in critical information literacy?

We may find it more instructive to ask students to start with reflection and analysis about: 

• what these tools can (and can’t!) help us to do,
• why we should (and shouldn’t) use them, and 
• how we can employ these tools in ethical ways to develop our own understanding. 

At OU Libraries, we’re providing targeted opportunities for students to engage with these kinds of critical thinking experiences around LLMs.

Student AI & Critical Information Literacy

WHAT students can - and can’t - expect from LLMs, information-wise

The Library’s micro-course, Introduction to AI Literacy, provides brief but targeted opportunities for students to explore what generative AI tools are and are not. We highlight the differences for students between generative AI tools and search engines (although these distinctions are becoming less apparent), and we emphasize what each tool may be most useful for. We also highlight the different roles that generative AI tools like ChatGPT, Gemini, or Claude might play, which provides students with new ways of considering what such resources can help them to learn and accomplish.

This resource is available to anyone at OU, so you might find it useful to assign if you’re dealing with AI Literacy in your course. Faculty may also find it useful to explore the what of generative AI with students, either for the LLMs we all encounter or for discipline-specific tools. Asking students to critically consider what such tools can do, in a particular context, can help them to engage more intentionally with these resources, as well as choosing when to not use such tools. Faculty may also want to ask their students to critically evaluate what information is being generated by these tools on their discipline, along with what is missing.

WHY students may - or may not - want to use LLMs to find information

When we consider the why of LLMs and information sources, it involves thinking about why we would – or wouldn’t – use these tools to gather information. Just as WRT 1060 provides an essential foundation in academic writing, the library-focused WRT 1060 Online Library Module explores what academic research is, what makes something an academic source, and how students can think about looking for, finding, and making sense of information sources. We’ve decided that generative AI tools have to be a part of these conversations, because these provide opportunities for us to help guide students to more critically consider why they might engage with LLMs, and why these tools may not be useful.

We highlight that LLMs – particularly OU-provided tools – are useful brainstorming resources in the research process. For example, we call out Gemini as a tool to scope out a potential research topic, figure out keywords or ideas we could search for, or give us a high-level overview we can use as a starting point. We also emphasize that LLMs are not useful sources to find academic articles because these resources are often behind paywalls and aren’t part of the LLMs or discoverable online. Indeed, we’ve seen many examples of fabricated citations generated by the predictive logic employed by generative AI tools 

In other fields, faculty may also find that intentionally defining the why – alongside the why not – can help students to become more critical consumers of LLMs, during their academic experiences, in their professional careers, and in their personal lives. Just as the Libraries have developed course-specific instruction in how LLMs impact information literacy, library faculty can work with disciplinary faculty to develop targeted and relevant guides, learning, modules, or in-class instruction to address these issues in their course content.

HOW students can use LLMs to find academic information

Examining the how of LLMs and information sources involves critically considering how we use these tools, in ethical ways, in looking for information, seeking to better understand ideas, and figuring out how to make sense of difficult-to-understand academic sources.

In the Libraries’ Academic Integrity in Research and Writing micro-course, we’ve added a new module on Academic Integrity and Generative AI, with an emphasis on academic research. Here, we’re able to highlight the built-in LLMs in library resources – like the AI-assisted search in Library OneSearch or the Research Assistant and AI Research Tool features in subject-specific databases (see the AI Integration in Library Resources guide). 

While these tools are not as expansive as ChatGPT, Gemini, or Claude, they are LLMs, and the boundaries around how these tools work can actually be helpful ways to begin with generative AI in academic coursework. The AI-assisted search option allows for natural language searches, which the LLM uses with Boolean logic to expand research while remaining focused. The Research Assistant and AI Research Tool are a step up: these options, built into subject-specific databases, can break down an academic article’s key ideas, identify connected concepts, and help make the complicated language of academic articles more understandable. 

For faculty in other subject areas, they may find that the generative AI options built into library resources are useful starting points to explore how students can (and can’t) employ LLMs in their learning. Because these options are more structured and controlled than open-ended chatbots, these may provide a “training wheels”-type option for students (and faculty) as they think critically about how LLMs impact their meaning-making. What’s more, faculty may find these tools useful for their own research.

And a revised SIFT Method for AI

In addition to these approaches that OU Libraries offer to our students, there’s also a revised SIFT Method that Mike Caulfield offers as a preliminary way to think about those information sources that are mediated by AI tools. In this model, SIFT stands for: 

• Stop – get information into a LLM about crafting the perfect prompt
• Using the LLM to investigate the evidence provided and gather context on the claim – while also
• Prompting the LLM to find better sources 
• And getting out of the LLM to track down sources from the LLM in other tools / information sources

These revisions emphasize the role that LLMs can play in helping students to develop critical context, which may be useful in some cases – particularly if we’re fact-checking a claim or trying to understand the veracity of an idea. 

Conclusion & Looking to the Future

While the SIFT Method may still be helpful for us to structure critically engaging with information resources, we can’t avoid the impact that LLMs have had on how we find, understand, use, and synthesize this information. But if we intentionally contextualize the roles that generative AI tools can – and can’t – play in this work, we can still foster critical thinking (and ignoring)... both in our students and in our own practices.

References and Resources

Caulfield, M. (2026, March 29). A SIFT Rebuild in Progress. From the blog The End(s) of Argument.

About the Author

Amanda Nichols Hess is a Professor and the Coordinator of Instruction and Research Help in the Kresge Library. This year, she is also serving as the CETL Faculty Fellow for AI. Her research focuses on information literacy instruction, faculty development, online learning, and how these concepts intersect. Outside of work, Amanda moonlights as her children’s elementary school mascot (which is a Dragon).

Others may share and adapt under Creative Commons License CC BY-NC.

Receive Weekly Tips View All Teaching Tips Submit a Teaching Tip