What are you working on right now, and how did you get into it?
Currently I’m leading a research group in the field of Human-Machine Interaction (HMI) with a special focus on voice assistance systems, in other words our Allinga activities. Our work focuses on human-centered research and development with three primary pillars: user research, design, and data governance. Right now, I’m evaluating large language models and the transparent design of natural language processing (NLP) systems. I chose these topics because I see a need for them. Industry representatives report that evaluation is difficult and that they don’t want to depend on leaderboards in their development work. On the one hand, leaderboards illustrate how good language models have become. However, there are also problems, such as testing data leaks. Data that is used for tests can flow into large language models as training data. It’s comparable to schoolchildren stealing test questions before an exam and memorizing the answers. They get an A grade, but they don’t actually learn anything. We’re studying how to make evaluation reliable and meaningful. Transparent design of NLPs is another topic I care about. European laws stipulate that users have a right to precise, comprehensible and accessible information on how their data is processed. This is currently not the case with applications like ChatGPT and Alexa. We’re working on systems to improve the state of the art so users have a fair chance of making informed decisions.
What motivated you to apply for TALENTA?
My supervisors encouraged me and supported me apply for TALENTA. I also liked that there’s a program devoted specifically to supporting women at Fraunhofer. I’m actively helping women in STEM subjects and professions, like in connection with Girls’ Day and through Femtec and mentoring. Not enough women dare to enter this field. I want to encourage them by setting an example. This also includes accepting offers of support myself. Seeking and accepting help is a soft skill, just like offering it. I learn constantly from my colleagues and am grateful for this.
What career goal were you pursuing by participating in TALENTA at Fraunhofer IIS, and what are your plans for the future?
When I participated in TALENTA, my goal was to become a group leader. That didn’t happen while I was in the program, but about a year and a half later, I was able to assume leadership of a group. It wouldn’t have been possible without the support of my supervisors. Going forward, I’m planning to develop my group further so we can show together that human-centered research and development create value. What was the main way in which TALENTA added value for you and your research work? The biggest benefit of TALENTA for me and my research work was the comprehensive support it provides. Thanks to this program, I now have access to a broad network of experts and like-minded individuals. I also received financial resources for targeted coaching, which has enabled me to deal with challenges even more effectively and learn new problem-solving strategies.
What motivates you to do research work, and what hurdles and advantages have you personally experienced over the course of your career in science?
In my daily work, I’m motivated by the exchange with colleagues and students; it’s a regular source of new ideas and perspectives for me. I find the topics we work on exciting and meaningful. This is also reflected in my group, where I see a lot of intrinsic motivation. During my university studies I didn’t believe that women were disadvantaged in science. But now, ten years and a child later, I see things differently. Disadvantage can be a hurdle. Fortunately, TALENTA and mentoring programs have helped me successfully pursue my career in science. I also experience the culture at my workplace as very family friendly. Was there a particular experience that sparked your passion for science? What drives me is a passion for people. Science takes second place. I have the impression that I can’t let go of science after years of training myself to think scientifically. Science has become ingrained in my thinking. In other words, science won’t let me go, and at the same time I’m driven by a passion for people. What opportunities are there in your research area to address current societal challenges? In our research field we’re focusing on AI language technologies. These technologies have been booming for years and changing our everyday lives. Take Alexa and ChatGPT, for example. We’re using human-centric evaluation and transparent data processing to address two very pressing social challenges that these technologies pose.
This interview was conducted by Saskia McDonagh, editorial team, Fraunhofer IIS Magazine.