Towards User Interface Design Guidelines for Effective Human-AI Interaction in Bioacoustics

Advancements in artificial intelligence (AI) have shifted how wildlife research programs analyze large acoustic datasets. Effective human-AI interaction (HAII) is required to maximize the value of AI and the domain expertise of human experts. However, to our knowledge, no guidelines for designing graphical user interfaces (GUIs) that enable HAII in bioacoustics exist. Accordingly, our study synthesizes bioacoustics and HAII literature to develop seven design guidelines supporting the design of GUIs for annotation and quality control. We demonstrate our guidelines by developing a prototype GUI and performing descriptive evaluation, as well as back testing of our guidelines against relevant extant instantiations. Overall, we find that our guidelines are well specified and suitable for their intended task. Our study represents an initial foray into the specification of guidelines for GUIs in bioacoustics and contributes to an important gap in knowledge and practice.