Back to the previous page
Jakub Mlynar
Monday 15 May 2023 08:00

Why does an IT research institute work with a social scientist? Jakub Mlynar, PhD in sociology and collaborator at the institute of Informatics of the HES-SO Valais-Wallis talks about his projects and research. Today, machines, algorithms and artificial intelligence are an integral part of our lives: smartphones and computers accompany our professional and private lives, our vehicles are electric and connected just as much as our refrigerators or watches. Our interactions with these new tools modify our human behavior and this is precisely the field of study of Jakub Mlynar who is interested in human-machine interactions within the HUCO - Human Centred Computing Group, the research laboratory of Professor Florian Evéquoz.

Conversation, a simple art, yet so difficult to learn.

Jakub Mlynar, originally from Prague, is happy to be able to work in Switzerland, especially at the HES-SO Valais-Wallis, because he can collaborate with people from different disciplines. He is looking forward to thinking about a common language between different scientific fields to look at the problem of interactions between technologies and humans with new perspectives. How do people use technology? This question was the basis of Jakub Mlynar's studies when he was a student at Charles University in the Czech Republic. He focuses on digital technologies and the daily interactions we have with them. Indeed, things that seem simple and natural to us, like having a conversation with another person, is an activity that requires a lot of effort, practice and is not easy to learn or explain. So, imagine how complicated it is to teach a machine - a conversational agent - how to make a conversation feel natural? It is these seemingly insignificant, yet very consequential, everyday human interactions that the researcher analyzes to be able to describe and explain what constitutes them (language, gestures, and non-verbal interactions). This should allow us to better adapt digital technologies to our behaviors to make human-machine interactions more fluid.

Technology modifies our perception of information and affects our behavior.

During his PhD in the Department of Sociology, Faculty of Arts at Charles University, he analyzed oral history interviews (available at the Malach Center for Visual History at the Institute of Formal and Applied Linguistics). In doing so, he found that technology can change our perception of stories that are recorded in the present to speak about the past to people in the future. The medium of video to capture these stories not only changes the way the protagonists tell their stories but also the way the stories are perceived by the audience. During his post-doctorate, Jakub added a dimension to his research by examining how these types of filmed interviews could benefit the learning of students by broadcasting them in classrooms.

It is in Switzerland that he continues his work with Esther González-Martínez at the University of Fribourg, between 2016 and 2018, thanks to a grant of excellence from the Swiss Confederation. He focuses his research on the interactions between the people present in a classroom rather than between the video and the people watching it. How do students work together using a computer and what information is retained during learning through the mediation of video? All these questions lead to improvements in the systems we use every day, such as video conferencing. It is important to know how these tools affect our social interactions and how information technology impacts the environment and human behavior. In the future, this will allow us not only to use more efficient systems, but also to devise legislative safeguards to control their deployment.

Artificial intelligence is more about acquiring skills than intelligence.

From September 2021, Jakub Mlynar will be working at the HES-SO Valais-Wallis, focusing on artificial intelligence as a phenomenon and a social object. Rather than the technology itself, he is interested in the changes it brings to the interactions of humans with each other, between the tool and the human, and how it is integrated into everyday life; think of automated robots in factories, autonomous vehicles or chatbots (conversational agents). As part of the Mobility Lab, he worked on the "Robi at Work" project in partnership with PostAuto and Benjamin Nanchen of the Institute of Tourism to find out how this luggage-carrying robot in Saas-Fee affected human behavior. He discovered that people on the street, the users of the robot, the people who recorded the videos of the interactions in the field and the teleoperator in charge of the semi-autonomous robot all participated to varying degrees in the movement and smooth running of the machine to ensure proper operation. It therefore takes a lot of people to make this type of tool work properly, since its autonomy is quite relative! The researcher finds it essential to bring a sociological and critical vision of the technology because if it is a source of comfort, it can also be at the base of the polarization which separates people on the social networks, of democratic deficit or of social inequalities for those who do not have access to it because of financial means. Before flooding the market with new technologies, it seems wiser to ask what impact they will have on humans, their interactions, and their environment. "We live in a time when we know resources are limited, so we need to ask ourselves if the technologies we are imagining meet a need. If we want to act responsibly, it is no longer possible to introduce new technologies to the market without asking ourselves what the social, human and resource costs are," says Jakub Mlynar.

Personalized medicine is possible if technologies are transparent.

In the medical field, information technologies such as artificial intelligence can complement the work of the medical profession. Indeed, QuantImage is a research platform that allows medical staff to easily build and evaluate AI models for diagnostic purposes, such as differentiating between a cancerous or malignant nodule based on an X-ray image. The medical profession is actively involved in this project, which could provide decisive elements to caregivers to guide the choice of an adequate treatment, but obviously remains cautious in front of tools to which it could delegate important responsibilities, including those of human life. Jakub Mlynar wanted to understand how to use the radiology platform designed and developed by Adrien Depeursinge's team, a researcher in eHealth at the institute of informatics, as efficiently as possible. To do so, he filmed about thirty caregivers in groups of two using the QuantImage platform to analyze radiological images. By analyzing the students' interactions with the software, the researcher was able to identify problems with the use of the platform and suggest improvements in design and functionality. All these researches show that we often interact socially with digital tools and machines as if they had a personality. These tools and machines are part of our society, and we need to think about their usefulness, the trust we place in them, the transparency of their operation or the level of responsibility we are willing to delegate to them. Geoffrey Hinton, nicknamed the "godfather of AI", recently left his position at Google to denounce the risks linked to the technology he helped develop. However, we must also be aware that artificial intelligence is not as intelligent as humans can be thanks to their mental and cognitive capacities. Artificial intelligence allows to perform specific tasks close to practical skills. It should be taught and explained in these terms to avoid confusion with human intelligence.