Application of Artificial Intelligence in Social Work Qualitative Research

Background and purpose: The ethical utilization of artificial intelligence (AI) tools in education is a topic of conversation worldwide across education disciplines (Crompton & Burke, 2023; Tahiru, 2021). Such technologies have been demonstrated to be beneficial for both education and research by assisting in tutoring, the development of evaluations and administrative tasks such as grading (Crompton & Burke, 2023; Tahiru, 2021); however, ethical considerations related to access to the technology and training, accountability and accuracy of AI systems and authenticity of authorship remain (Tahiru, 2021; Elaili & Rachid, 2023). While the use of generative AI has been at the forefront of controversy (Wach et al., 2023), student utilization of any form of AI in research and education carries implications for higher education policy. Therefore, this poster describes the process of comparing methods of transcription to justify the use of an AI transcription service (OtterAi®) for the qualitative portion of a dissertation.

Methods: Three qualitative interviews were transcribed initially via Zoom® transcription service. To compare the efficacy of Zoom® transcription with OtterAi®, with written consent from research participants, audio recordings of the dissertation interviews were uploaded to the OtterAi® transcription service. Transcripts were compared side-by-side for accuracy and formatting. User experiences utilizing both OtterAi® and Zoom® transcription are reported. Side-by-side comparisons of the transcriptions will be displayed within the poster presentation.

Results: Compared to Zoom, OtterAi more accurately grouped together the text from the same speaker, requiring fewer formatting changes. Likewise, OtterAi recorded fewer “ums” and filler words. OtterAi allows researchers to listen to the interview audio within the OtterAi interface in real time. While both OtterAi and Zoom allow rewinding and fast-forwarding, OtterAi also allows users to start and stop playing audio by toggling to the specific section of interest in the document itself. OtterAi also allows users to make changes to the transcript in real time without navigating between Zoom and another word processer. Thus, editing can be done much more easily and efficiently within the interface. OtterAi also provides a button which allows users to easily change the speaker when text was not grouped together. Moreover, OtterAi allows users to search within the document, highlight specific parts of the transcript and add notes, which can support initial qualitative analysis. OtterAi transcription saved approximately three hours compared to transcription with Zoom.

Conclusion: OtterAi demonstrated considerably greater transcription accuracy and usability when compared to Zoom transcripts. Therefore, when considering policies related to the use of AI within higher education, policy makers should account for nuances in the application of AI, such as the use of transcription services, when making policy statements. Thus, the use of AI-powered transcription services is an important innovation holding positive implications for qualitative research.