Researching with GenAI: Tools & Tips for Special Librarians

When generative artificial intelligence (GenAI) first became widely available, there was a great deal of discussion about how it would “hallucinate”—that is, make up content and sound very sure of itself when doing so.

From my personal experience, if I asked GenAI to provide references it would comply—but they would rarely be actual sources. Instead, the AI would give me references to sources that sounded very legitimate and appeared to be by authored by real people. But when I dug deeper and tried to find the original, it became apparent that the source did not exist. In other words, the AI had “hallucinated” a source of its own creation, rather than providing me with an authentic one.

More recently, though, I’ve found that GenAI is getting much better at locating sources that do exist and are directly related to the topic I’m researching. Even with the free version of ChatGPT, for example, I have asked for references for a topic and it has accurately provided them. However, if I ask for 10 references, it may only provide 4 unique ones.

It’s certainly not perfect, but this growth of GenAI’s abilities encouraged me to start experimenting with it for research. As librarians, our abilities to locate sources are greater than most. So, while I still haven’t found GenAI to provide new sources that I would have been unable to locate myself using traditional databases or Google Scholar, I do see how GenAI could benefit anyone less familiar with literature searching or less familiar with a specific topic.

7 GenAI Tools for Special Librarians to Explore

Now and in the future, GenAI tools are going to affect the research process. In particular, the following GenAI tools may be worth familiarizing yourself with (I included taglines to provide context for each tool):

• Elicit: “Analyze research papers at superhuman speed.”
• Consensus: “Find the best science, faster.”
• Research Rabbit: “We’re rethinking everything: literature search, alerts, and more”
• Semantic Scholar: “A free, AI-powered research tool for scientific literature.”
• Scite: “Enhance your research with an AI built for trust.”
• Connected Papers: “Explore academic papers in a visual graph.”
• Perplexity: “Where knowledge begins”

These tools provide unique ways of conducting research and visualizing the results. I encourage you to explore them and determine how they may be beneficial for your own research or the research your stakeholders are conducting.

AI Specialized for Research vs. General AI Tools

While there are unique AI tools for research, I do think the more standard GenAI tools such as ChatGPT, CoPilot, Gemini, and Claude will continue to get better at research. Even now, ChatGPT has ways to add data visualization. It will be important for us to know how to use these tools, provide support to our stakeholders who can benefit from them, and be able to explain the drawbacks and concerns GenAI introduces into the research process.

More on the drawbacks and concerns in an upcoming post…