Center for Gender and Diversity Research

International Online Conference "Fair Medicine and Artificial Intelligence"

From 3rd to 5th March 2021, the international online conference "Fair Medicine and Artificial Intelligence: Chances, Challenges, Consequences" will take place.

We invite scholars from the social sciences and related disciplines like philosophy, medical ethics, and public health research who engage with questions of AI in medicine and the health sector to submit an abstract.
We welcome abstracts and papers in both English and German.

The Call for Papers can be found here.

For further questions and information, please contact Dr Renate Baumgartner  medAI.conference2021spam prevention@gmail.com

Excerpt from the Call for Papers

At least since 2012, and following technological advancements in IT, the medical profession has become increasingly interested in artificial intelligence (AI). An aging society, the need to balance rising costs in the health sector with a certain stability in the average health of the population while trying to keep health inequalities in check have all contributed to investing AI with the hope to enable more successful medical care and better health for all. Visions range from seeing AI as a universal remedy, able to solve the key challenges of contemporary medicine, to the dystopia of a health care system without human medical staff. Medical diagnosis, prognosis (e.g. in personalised health care), and therapy recommendations are all possible application fields of AI, to name but a few. Despite the high hopes for AI in the field of medicine, only a few products have so far managed to meet the standards necessary for broad marketability in terms of adequate available data or validation. Even regardless of the velocity of developments, AI will most likely play an important role in the health sector in the near future. Healthcare disparities are posing a political threat and a major challenge to the healthcare system. The use of AI in the service of fair healthcare makes for a persuasive argument that not only justifies its employment, but seems to make it more or less inevitable. AI could, for instance, reveal human bias in the field, and make equal treatment available to all. On the other hand, critical voices warn that AI might heighten existing inequalities, while technical complexities would make them harder to detect. The question is thus whether algorithm-based applications can influence systemic inequality in positive ways