Developing a Typology of Women’s Attitudes to AI Use in the Australian Breastscreen Program – a Qualitative Investigation of Attitude Types and Perceived AI Acceptance.

Applications of artificial intelligence (AI) in the healthcare sector have been one of the most prominent growth areas in the AI industry. Radiology is a natural nidus for this growth given it’s pre-existing digital nature, large well curated data sets and quality assurance. Scientific evidence supports AI use in breast screening representing a tremendous opportunity to improve the accuracy, experience and efficiency of mammography reading. While such innovative approach holds promise, there is uncertainty around how it can be successfully integrated in the breast screening services whilst maintaining faith and participation from current clients. Existing empirical studies investigating lay women’s perceptions to AI use in breast screening highlight positive and negative attitudes towards AI and AI use. This study develops a typology of women’s attitudes to AI use in a breast screening service, which is supported by the work of Birkland (2019) that has developed the ICT user typology among older adults, and investigates a relationship between attitude types and levels of AI acceptance. We conducted a combination of focus groups and one-on-one interviews with the clients of BreastScreen program in Australia. In total 26 clients participated. We found the participants fall into four attitude types – “Enthusiast”, “Practicalist”, “Traditionalist” and “Guardian”. The majority was categorised either Enthusiast or Practicalist, showing a high or certain level of AI acceptance. They even noted AI integration into the existing BreastScreen program would enhance their trust in it. Two fell in Traditionalist and other two fell in Guardian demonstrating AI resistance. Reasons and motivations that ascribe each woman to the certain category varied individually based on their lived experiences, and they also had a significant impact on their perceived AI acceptance and unacceptance. Practical implications based on our findings will be further discussed.