Generative AI tools can expedite and enhance several of the processes involved in creating a literature review. They can also fail in ways that are easy to miss. Source summaries can misrepresent the content; syntheses can obscure important nuance; and analyses can miss key insights presented in figures. Unfortunately, the more polished the output looks, the more trust it gets. This session addresses how to use these tools effectively while being realistic about what they can and cannot do. 

We'll start by mapping the current landscape, from general chatbots like Claude and ChatGPT to dedicated scholarly tools like Elicit, Consensus, Undermind, SciSpace, and Asta, to new features in traditional library databases. You'll learn how to craft prompts that leverage the affordances of semantic search. We'll also examine where these tools tend to go wrong, both in finding sources and in describing what those sources actually say. And you'll leave with concrete strategies for checking AI output. 

This session is designed for faculty and graduate students who have some experience with generative AI and want to use it for serious research. No coding required. It is the first in a series. Later sessions will go hands-on with the dedicated tools and show how general chatbots can be turned into research agents. It is organized by the K-State AI Consulting Service (KAICS).  Note: The session will be recorded and shared with all registered participants.

Register to attend