AInterviewer: a framework for using Generative AI to collect large-scale qualitative interview data

Large language models (LLMs) are increasingly used in social sciences to classify content or as synthetic 'respondents', but their greatest potential may lie in eliciting new information from people. We ask the following research question: is it possible to leverage generative AI to conduct high-quality, ethically responsible qualitative interviews on a large scale? We propose the AInterviewer: a novel framework for collecting large-scale interview data, based on the application of cutting-edge LLM techniques in the social science domain. AInterviewer will be implemented and tested to ensure ethically responsible and high-quality interviews while being GDPR compliant. We will apply the AInterviewer to a case study of Danish citizens' opinions towards the green transition to produce large-scale, in-depth data to inform better data-driven policies.