Introduction The adoption of generative AI (genAI) tools like ChatGPT present opportunities for enhancing efficiency through automation. While genAI use is increasingly commonplace in academia and research, the capabilities and limitations of such tools to support systematic review (SR) searches has yet to be explored. Methods We recruited Canadian health librarians to participate in a two-stage crossover study followed by interviews to elicit librarian perceptions of using ChatGPT for SR searching. Each librarian developed two Ovid MEDLINE searches, one using ChatGPT and one without. ChatGPT training was provided to participants when they were required to use ChatGPT. Interview transcripts were coded for themes using qualitative content analysis by three independent coders. Results Eight of the ten participants preferred traditional searching methods over the use of ChatGPT. Themes that emerged through inductive coding of the transcripts include: Librarian Preferences, Barriers to Adoption and Risks, Value and Strengths of ChatGPT, Adoption Bridging, and Positioning AI within Librarian Practice. Discussion This study provides practical examples of how health librarians may use ChatGPT to expedite their search development process. It also revealed the potential and key limitations of ChatGPT use in its current form. This study highlights the importance of staying informed on the development of GenAI tools.