A Research Assistant with No Relations: Indigenous Insights on AI in Data Analysis

Abstract

There has been growing ethical concern about the use of Artificial Intelligence (AI) in research, as well as calls for its use to increase efficiency in qualitative research. However, we must consider the impacts on our research when we outsource critical tasks to a tool with no responsibility, experiences, or relationships. The pro-AI argument that its use will allow qualitative scholars to efficiently sift through enormous amounts of data promptly does not consider the meaning-making involved in the tasks delegated to the AI, which may reduce the effectiveness of research outcomes. Others have addressed these drawbacks by curating datasets that center participants' values and co-creation of the datasets with the communities involved in the research. These actions overlook the crucial aspect of researcher positionality—the relationships, responsibilities, and experiences the researcher has with their research. Indigenous research methodologies emphasize the contextuality of research and the meaning-making role of the researcher when interpreting data. Self-location is a critical tool in Indigenous research, allowing researchers to reveal the experiences, relationships, and responsibilities that they have to the research, which influences the questions they ask and how they interpret data. AI systems lack positionality and therefore cannot consistently ground their interpretations of data—they decide which data are relevant based on opaque algorithms. We recommend that researchers reflect on their own positionality through the practice of self-location before engaging in research and deciding whether to use AI, and whether doing so will add value or have epistemic drawbacks.