30. March 2022

#DiversityInScience - The Feminine Face of Artificial Intelligence

Humans and artificial intelligence (AI) are the team-mates of the future. One approach taken to achieve cooperation that works well is “human-centred artificial intelligence (AI)”. Explainability and trust are essential aspects here. In the ongoing development of AI, women are having an ever greater say. Female researchers from the UAR Innovation Network give insights into the research questions that point the way forward for “AI team-building” and explain why diversity plays an important role in research.

Close connections in the AI community
Within the UAR Innovation Network, intensive research in the field of human-centred AI, such as that conducted at the two Hagenberg-based research centres RISC Software and Software Competence Center Hagenberg (SCCH), is a central objective of Upper Austria’s business and research strategy #upperVISION2030. The centres maintain close relations in the AI community in other regions. SCCH research director Bernhard Moser is president of the Austrian Society for Artificial Intelligence (ASAI), the platform for Austria’s AI community. The research organizations are also members of the nationwide network AI Austria and highly committed members of the regional association AI Upper Austria of which SCCH is a founder member. Through this local group, the international community is strongly represented in the industrial region of Upper Austria. The association Women in AI ensures that women have a big say in the field of AI. This network of female experts brings together woman researchers from the centres with the specific aim of promoting diversity and raising the profile of women in AI.


Promotion of women in research
The research centres carry out initiatives with the specific objective of promoting women and young talents. These initiatives are often the result of the efforts of ambitious women to whom they are very important. Overall, the gender gap in research has narrowed in recent years, particularly among young researchers. At the research centres SCCH and RISC Software, women account for one-third of the team on average. This share is considerably higher than the approximately 20% in the IT branch overall. The intention is to cement this strong position still further. Both centres maintain partnerships with technical secondary schools, and so take research on topics relating to the digital transformation and artificial intelligence into the classroom. The centres also regularly participate in exciting school competitions in order get young people interested in technology, and fascinated by it, from an early age.

What role does explainability play in AI models?

“Part of the great potential that learning algorithms have is the ability to learn from complex data and exploit previously unknown relationships. Explainable AI can help to uncover these relationships, understand them and so generate new knowledge. In this way, explainability can help to identify any weaknesses in an AI model (e.g. data bias) at an early stage and prevent undesirable consequences such as discriminatory decisions made by the algorithm. At the same time, this also creates greater trust in AI methods, which is a fundamental prerequisite for successful cooperation between humans and artificial intelligence.”

MANUELA GEISS is senior researcher for data science at Software Competence Center Hagenberg. She focuses particularly on deep learning and explainable AI (XAI).

Why is gender balance important in AI research?

“Machine learning and artificial intelligence open up new opportunities. However, they also pose challenges. AI systems can automatically make discriminatory decisions if the data sets used reflect prejudices. This can be prevented by teams of experts with a good balance since different, gender-specific perspectives can be incorporated in research work early on. The wide range of applications for AI systems requires not only specialist technical expertise but also knowledge that can help to resolve the numerous ethical and legal questions.”

ROXANA HOLOM is a data science project manager & researcher at RISC Software. The Romania-born scientist has been an active member of the Women in AI Upper Austria platform since the year it was founded.

How can more women become involved in research?

“There is a highly creative side to research. Imaginative ideas and a fondness for resolving a wide variety of challenges are needed. That’s why it is important to question subconscious prejudices and stereotypical role models and get young women interested in STEM professions. Work experience aimed specifically at girl pupils and woman students provides valuable insights into the work environment. In addition, training courses for female researchers are to be intensified and mentoring programmes launched with a view to improving career opportunities for women in research. At the same time, general conditions must be created so that, as in every type of occupation, family life and working life can be perfectly balanced. Such conditions include part-time work models, opportunities for returning to work following parental leave and work-family balance models.”


VERENA GEIST is key researcher, software science at Software Competence Center Hagenberg. She studied software engineering for medicine at the University of Applied Sciences Upper Austria and completed her doctorate sub auspiciis at the JKU.

How does diversity benefit research?

“Diversity is an essential component that must be reflected in research. Press reports during the past year have shown what can happen if diversity is ignored in research, from the gender gap (in medicine, for example) to discriminatory algorithms. This is why it is particularly important that research takes account of individual circumstances such as gender, age, disabilities, different cultures and much more besides. It can do this through the research topics it chooses as well as through project teams that make this diversity an integral part of their daily activities.”

CHRISTINA HOCHLEITNER is a research coordinator at RISC Software. She has already been chosen expert of the month by the nationwide FEMtech initiative in recognition of her efforts for the promotion of women.

Does research need more female role models?

“Role models that motivate and inspire can be a powerful influence when choosing a career and in other areas too. Gender, as one aspect among many, can play a significant role when trying to reach girls and women. What I find crucial is that female role models show us what we could be in future and, at the same time, remind us a little of ourselves. For this, we must be able to classify them as ‘someone like me’.”

ANNA-CHRISTINA GLOCK is a researcher and data scientist at Software Competence Center Hagenberg. She was actively involved in the “Girls! Tech up Role-Model Award”, focusing on encouraging more girls to choose careers in technological disciplines.

What is the language of young research?

“Unlike some of my colleagues I grew up with the internet. My first encounter with a floppy disc was as the ‘save’ symbol in Word. Only later did I discover that the thing actually exists. I think that the more you come into contact with technology as you grow up the fewer inhibitions you have about using it for everything. You’re not afraid about the third wearable that monitors your every move all day. You don’t panic at the thought that robots might take over only because they support our everyday activities. In my view, young research is dominated mainly by a rather naive approach. More openness and less fear of possible consequences because we can grow up in great freedom.”

ANNA-SOPHIE JAEGER joined RISC Software following a period of work experience in the field of software development that was arranged through a partnership with her school. After graduating with a BSc in medicine and bioinformatics she is now studying data science and engineering.

How does artificial intelligence understand human language?

“Natural language processing is an interdisciplinary field of linguistics, computer science and artificial intelligence that enables computers to read, decipher and understand human language. The availability of increased computing power, enormous amounts of data (big data) and modern algorithms continually leads to numerous revolutionary accomplishments. What is now needed are experts capable of achieving the leap from the research environment to production.”

SANDRA WARTNER is a data scientist at RISC Software and works on a variety of research projects and projects for customers. She focuses on a wide variety of tasks in data analytics (in the field of natural language processing) and on implementing AI solutions in practice.

What does diversity have to do with data quality?

“AI systems make decisions based on the data that was used to train and test the algorithms. If this data only reflects the perspectives of one particular group, AI algorithms will be unable to assess the realities of other groups of people. This effect is known as bias. To make objective decisions, data of sufficient quality is therefore needed. Among the parameters examined are completeness, correctness and currency. Attention must be paid to diversity not only when developing AI algorithms but also when checking the quality of the data: only if a wide variety of views are represented in the team can an inadvertent bias in AI-based decisions be prevented.”

LISA EHRLINGER is senior researcher, data science at Software Competence Center Hagenberg and conducts research in the fields of data management and data quality with a focus on “big data processing”.

Diversity in Practice in 6G Research

The UAR Innovation Network consists of a total of 17 outstanding research institutions that provide businesses with practical support for their innovation strategies. Their core competences can be summarized as three specialist fields: smart systems, digital technologies and sustainable materials.