Thursday 28th November 2024, 11:15–11:30 (Australia/Melbourne), Sutherland Room
The integration of Generative AI to support researchers analysing free text responses from public consultations presents an opportunity for rapid and high-quality synthesis for Government. By automating labour-intensive tasks with easily defined rules, such as thematic coding, researchers can focus their time on delivering high-quality qualitative research and critical analysis. However, this approach raises important considerations regarding ethics, privacy, transparency, algorithmic bias, reproducibility, consent, validation, and data quality.
The Social Research Centre uses a range of Gen AI approaches. Using a recent project involving public consultation on a draft cat management strategy delivered for the Victorian Government. This project involved the analysis of 70+ survey questions and 85 written submissions. Generative AI was utilised in the qualitative analysis of 11 free text questions totalling over 8000 responses, in addition to the 85 written submissions totalling 499 pages.
We will discuss the benefits and drawbacks of different Generative AI approaches taken at various stages of the project, sharing valuable learnings gained from these experiences. The presentation will highlight how we navigated the challenges associated with maintaining high standards of security, quality, transparency, and ethics while leveraging the power of AI technologies such as Large Language Models (LLMs).
I am a research assistant in the Evaluations and Qualitative Research team at the Social Research Centre (SRC). Since ChatGPT3 the AI field has absolutely captivated me, with every new model release I'm stunned all over again. I take a lot of joy in keeping up to date with the industry and have been fortunate enough to explore the inclusion of AI in SRC workflows.