Ute Schmid

Women are significantly underrepresented in the fields of computer science and AI. To what extent does this affect technology development?

Ute Schmid: If there are too few female specialists, there is simply a shortage of skilled workers. In the field of AI in particular - even more so than in other areas of computer science - there are currently more jobs on offer than suitably qualified applications. Unfortunately, girls and young women in particular still have fewer opportunities to discover whether they have talent for and interest in a subject like computer science. They are often less confident about studying computer science than boys - despite comparable academic performance, especially in mathematics. Empirical studies suggest that if women choose computer science, they are more likely than men to care that what they research and develop leads to helpful applications. Jane Margolis and Allan Fisher of Carnegie-Mellon University characterize the difference as "computing with a purpose" versus "dreaming in code." Without being able to back this up with concrete numbers, more women are found in AI application areas such as medicine. Women are also engaged in AI solutions for sustainability and are more active in developments for interactive AI systems than in fully autonomous approaches.

For another reason, it is important that women are more represented in AI development: Looking specifically at data-intensive AI - i.e., the field of machine learning - initial scientific publications suggest that diverse development teams can help reduce unwanted biases. That is, diverse teams pay more attention to ensuring that genders and ethnicities are more fairly represented, even in the selection of training data.

How can more girls and women get excited about AI?

Ute Schmid: From my many years of experience with offerings for girls in the area of computer science, I take away: The most important thing is to enable girls again and again, preferably starting at kindergarten age, to have very concrete experiences - be it through programming robots or interactive games, through applications of machine learning to classify images or sensor data. This should involve choosing topics that are more likely to appeal to girls but are not necessarily stereotypical. A classification of bird images is probably more exciting to them than a classification of car brands. Do-it-yourself activities not only arouse interest, but also give girls the chance to build up confidence in their own skills. It is also important that such programs are offered as continuously as possible so that the topic remains present as an option for their own study or career decision. Fortunately, my team at the University of Bamberg now includes several female doctoral students. They are not only doing excellent research, for example for partnered AI systems in medicine, but are also involved as role models for female students. I am optimistic that the proportion of women will increase in the coming years, especially in the field of AI.

What role models are there for girls and women in AI?

Ute Schmid: Currently, there are certainly fewer women in professorships than men. But some are very visible. - For example, the AI professor Manuela Veloso. She is a pioneer in the field of planning and learning, was very successful in the field of robot soccer for many years, and is president of the Association for the Advancement of AI (AAAI). Stanford professor Fei-Fei Li is extremely successful in the field of deep learning and computer vision. She focuses on applications in the field of health. At the same time, she is committed to inclusion and diversity in AI education through her AI4ALL initiative. In her publication "Gender Shades," AI researcher Timnit Gebru demonstrated that automatic face recognition models built with machine learning and used in commercial systems perform significantly worse for people with non-white skin color and females. There are also numerous female professors in the field of AI in Germany, many of whom are active with the Learning Systems Platform and are committed to sustainable, fair and human-centered AI applications.

Comprehensible information on the basics, applications and challenges of AI is provided by the web offering www.ki-konkret.de (in German) of Plattform Lernende Systeme. 

The AI map of Plattform Lernende Systeme shows which universities in Germany offer courses in Artificial Intelligence and data science.