Sowohl hinsichtlich der Daten, mit denen KI-Algorithmen „gefüttert“ werden, als auch hinsichtlich der KI-Entwickler selbst sind Frauen auf dem Gebiet der künstlichen Intelligenz unterrepräsentiert. Das sheHealth-Netzwerk möchte, dass sich dies ändert. Vertreter der Medien sind herzlich eingeladen, an der Veranstaltung teilzunehmen und darüber zu berichten.
The majority of participants in medical studies – for example, in the approval of new active substances – are white, middle-aged men. “There has long been a problem with study databases in medicine,” says Brigitte Strahwald, physician and epidemiologist at LMU Munich and a specialist in health data. This applies no less to AI research, which requires large amounts of data. “Not all data records are available in the required quality,” criticizes Strahwald.
Only one in five AI professionals is female
The situation is similar in AI research. Recently, the professional networking service LinkedIn found out in a study conducted for the World Economic Forum’s Global Gender Gap Report that not even every fourth AI professional worldwide is a woman. In Germany, the number of female developers is even lower. Professor Sylvia Thun, Director of the Core Unit eHealth and Interoperability at the Berlin Institute of Health (BIH), funded by Stiftung Charité, says: “We need more female medical IT professionals and female health data scientists; women definitely have some catching up to do here.”
The algorithm does not discriminate
A gender or ethnic bias when formulating study questions, the choice of applications and algorithms, or the development of interfaces between machines and humans has already produced unwanted side effects. It turned out, for example, that skin cancer screenings were sometimes only trained on white skin; voice recognition systems often have problems with female voices; the facial recognition function on smartphones has problems with women and people with dark skin; and recruiting tools display a preference for male applicants. Algorithms are based on criteria specified by people. “The algorithm does not discriminate,” explains Brigitte Strahwald. “Its performance is down to how the software is trained and what data it is fed with.”
Together with President of the German Medical Women’s Association (DÄB) Dr. Christiane Groß, Professor Sylvia Thun founded the network sheHealth with the aim of making the work carried out by women in the field of digital medicine and health more visible. Together, women from different backgrounds are working to raise awareness of gender issues and create greater recognition for female leaders and speakers. In order to advance the discussion about equal opportunities in AI research, the sheHealth network is now organizing a conference on women and artificial intelligence in health care at the Berlin Institute of Health.