Fraunhofer Institute for Digital Medicine MEVISInstitute News
Great hopes, but also weighty concerns: Currently, there is a lively debate about algorithms that can learn, especially in medicine where AI systems are supposed to help detect diseases and identify optimal future therapies. This raises important questions: How do the algorithms arrive at their results and how can they be verified? Who develops these new technologies that affect the health of so many diverse people? And who creates the stories and images that represent the new digital solutions in medicine? The Scottish-based creator Fiona Smith is exploring this field of tension with an elaborate art project that she will develop as part of the STEAM Imaging V residency program. Her idea: In her interactive installation “The Box”, the audience will feed an AI with data and then be able to observe how the machine reacts in an audio-visual display.
The Fraunhofer Institute for Digital Medicine MEVIS is hosting the “STEAM Imaging” residency program for the fifth time to critically examine topics in digital medicine. It enables young creatives to realize a project at the interface of art, technology, and science together with MEVIS experts. Within this collaborative learning process, students from the partner school SZ Walle are involved in a jointly developed STEAM workshop.
“New technologies and data-driven tools such as AI are changing the role of doctors and patients and our view of health,” says Bianka Hofmann, Head of Science Communication at MEVIS and developer of the residency program. “At the same time, the new tools are significantly shaped by the stakeholders in the development process as well as the selection of the underlying data.” That is why MEVIS is cooperating with the Institute of Design Informatics (IDI) at the University of Edinburgh for STEAM Imaging V. Among other things, it is dedicated to the question of how humans can effectively handle and interact with data streams in a wide variety of areas. “Design Informatics, with its connection to the Edinburgh Future Institute, allows for experimental and critical practice, informed equally by research and industry. Students build, test, hack and reinterpret what data means for society and where design has the opportunity to shape future thinking to realize speculative, forward-looking projects,” highlights IDI manager Nicola Osborne. “This makes the MEVIS residency program an excellent fit with our activities.”
The deadline for this call for proposals was the end of June. For the first time, it was addressed not only to artists, but to all master's students and doctoral candidates of the participating partner university. The choice fell on the proposal of a young medical doctor: during the COVID pandemic, Fiona Smith worked as a doctor in a hospital. “It became increasingly clear to me how strongly artificial intelligence influences my everyday life and will likely influence healthcare and my work as a doctor,” she says. She decided to pursue a doctorate in computer science: “I wanted to find out more about it and looked for a doctoral thesis on the topic of AI.” Specifically, it's about algorithms that could one day automatically monitor patients after they have had their operations.
Smith has used artistic approaches and processes to illustrate complex issues in the past. Among other things, she used her drawing talent to paint the bodies of volunteers for an organ donation campaign. “When I received the circular email with the call for proposals for STEAM Imaging V, it blew me away,” says Smith – and soon after designed a coherent concept. “The public often sees AI systems as a black box – no one really knows what's going on inside and how the AI comes to its decisions,” she explains. She wants to take up this idea with her interactive art installation “The Box”.
In the center is a two-meter-tall, mysterious-looking cube, equipped with a data entry station. Here, people can enter medical data sets and feed the AI with training data. Afterwards, the machine will use artistically enhanced data projection in the exhibition space to show what it has made of the different inputs. “If we feed the AI with incomplete or biased data – can we expect it to come up with reasonable solutions?” asks Smith. What kind of data sets specifically come into question and how the machine should present the results – these are the issues Fiona Smith wants to address together with the Fraunhofer experts during the two-week residency at MEVIS in Bremen in November.
“I'm really looking forward to working together, it's going to be intense conversations,” says Smith – and smiles. “I expect to be overwhelmed with data and information and will need to take a few weeks afterwards to digest it all.” After the intense weeks on site at MEVIS, the collaboration will continue in online meetings until the work is completed in the spring. She is equally excited about the two-day STEAM workshop that she will be running together with MEVIS specialists at the Walle school center. “It should be interesting to show the students the things I am working on to create the work, relate it to the research at MEVIS and see how they respond,” says Smith.
Afterwards, she will complete “The Box” to finally present it at the renowned Edinburgh Science Festival in April 2024 – at a special IDI exhibition venue: the “Inspace Gallery” offers ideal conditions for the presentation of elaborate art installations, but also allows passers-by to participate in the action by means of a huge outdoor projection screen. The exhibition will be accompanied by a discussion event at the opening and further STEAM workshops.
STEAM Imaging V is a creator residency of the Fraunhofer Institute for Digital Medicine MEVIS in Bremen, Germany, in collaboration with the International Fraunhofer Talent School Bremen, the Walle School Center in Bremen and the Institute for Design Informatics at the University of Edinburgh in the UK. It is supported by Ars Electronica in Linz, Austria.