Veranstaltungen 15. Wissenschaftliche Tagung am 20. und 21. Juni 2024

bstrac"Datenerhebung, Datenqualität und Datenethik in Zeiten von künstlicher Intelligenz

Privacy considerations for using AI in qualitative data analysis software

Dr. Jannine van de Maat

Netherlands Institute for Social Research (SCP)


Researchers within the Netherlands Institute for Social Research use qualitative data analysis software (like ATLAS.ti or MAXQDA) to analyze qualitative data. These qualitative data are collected via focus groups or interviews (or observations) by researchers from the Netherlands Institute for Social Research or by their contractors. The qualitative data analysis software tools offer AI plug-ins to help researchers with (automatically) coding their data, which can be a time consuming and subjective process. Full-automatic coding “powered by OpenAI” seems to be a valuable addition to the toolbox of qualitative researchers. But what are the drawbacks and considerations to take into account?

The Netherlands Institute for Social Research is a government agency and currently, the Dutch government prohibits the use of AI for all civil servants. In the long term, however, some guidelines will be developed as to what is deemed safe and useful use of AI.

Which privacy and ethical considerations should be taken into account in deciding whether using AI plug-ins is allowed for qualitative data? First and foremost are data protection concerns about the analysis of transcripts which are anonymized, but still contain information which may indirectly result in disclosure risks. Secondly, sharing confidential information (like concept publications) in AI tools might be undesirable. Are there other concerns and how can the risks for the use of personal data and confidential information be mitigated? And to what extent do the benefits of using AI plug-ins in qualitative data analysis software outweigh the risks? This presentation will reflect the current situation, including guidelines from the Dutch Data Protection Authority, the Dutch government and discussions in the scientific world. The aim is not to come to a final conclusion, but to review the considerations and reflections in a rapidly developing field.