News – 5 October 2022
With artificial intelligence, static products become dynamic; they change as they are used. This also requires new design methods. The European network and PhD program DCODE, coordinated by the Faculty of Industrial Design Engineering at TU Delft, explores the future of design in a world full of artificial intelligence.
What is a car? Everyone knows that. But the answer to this question is less obvious today than it used to be. A car in the digital age is fundamentally different from a car in the analog or mechanical age. In the latter, the car was an immutable product, designed for one driver and a number of passengers. In the digital age, the car is no longer an immutable product because it is connected, for example, to a navigation service, an image recognition system, to other road users and to road infrastructure. The digital transformation in general and artificial intelligence (AI) in particular turns a metal box on four wheels into a connected high-tech device that even changes over time.
Elisa Giaccardi, professor of post-industrial design at the Faculty of Industrial Design Engineering at TU Delft, gives this digitally connected car as an example of how AI forces designers to think in conceptually new ways.
The biggest change AI brings to design is that it enables systems to learn and evolve in and through use. What the product is keeps changing. Suddenly, a car can even become part of an insurance service. For us as designers, this means designing for new, developing relationships instead of designing a fixed product.
Prof. Dr. Dr. Elisa Giaccardi
How exactly design will change under the influence of AI merging with products and services is the big open question. To answer this question, Giaccardi coordinates the European network and PhD program DCODE. Within DCODE, fifteen PhD students in eight European cities are working from different angles on the same overall design challenge (for more information on DCODE, see the box). “We are trying to develop new design methods and a new vocabulary that will allow designers to communicate with each other about the new challenges,” says Giaccardi.
An example of a new concept in the design vocabulary is ‘multi-intentionality’. “Take a platform like Facebook,” explains Giaccardi. “Facebook serves multiple purposes and multiple interests at once, depending on who you are and how you interact with the platform. It serves different purposes, not all equally transparent, which is why we call it ‘multi-intentional’. The design universe is expanding, as networked products and services become multi-purpose through the integration of AI.”
Grace Turtle is one of the PhD students working within the DCODE program at TU Delft. She embarked on a project titled Queering AI in 2021. Turtle has both Colombian and Australian roots, considers herself queer and is inspired by her many identities without reducing herself to a fixed identity. Turtle: “For me, queerness as applied to design offers a way to explore the in-between. The in-between is fluid and resists categorization and normalization. Queerness asks questions and wants to be wild. Queering AI makes us rethink what AI is as coded, and how these encodings shape particular realities.”
Turtle illustrates his thinking with the way Google Maps works: “Google Maps tells you the best route from A to B. It interprets the term ‘best’ as the fastest route. Although the application gives you multiple routes, the most important measure is time, it takes to get from A to B. But when you apply the concept of queering AI, there should be several metrics to help you plan your journey.”
Maybe you don’t care about travel time and want to take the shortest route, maybe you want to pass by a specific point of interest, or maybe you want a little variety every time you travel from A to B.
Before starting her PhD, Turtle spent several years working in industry researching the future of cities. Turtle hopes that DCODE will help avoid the one-dimensional results that Google Maps provides. In her own work, she aims to expand the way we imagine the future of cities. Turtle: “At the end of my PhD project, I hope, among other things, to have created a queer-intelligent digital twin that opens up new possibilities for how we design, test and manage complex systems like a city.”
“The models and simulations that digital twins produce can be compelling decision-making tools,” Turtle continues. “Where traditional design processes lead to static decision-making or fixed outcomes, simulations allow you to explore ‘what if’ questions. This way you can explore in detail how a city can change. I want to explore what happens when we use the data, the model, and revisit the simulation through the lens of queerness.”
An important part of DCODE are the so-called prototeams. “A prototeam is a team of PhD students from different backgrounds who work together in realistic environments,” explains Professor Elisa Giaccardi. “Prototeams learn by doing and practice future career profiles.”
An example is the collaboration between one of the proto-teams with the Amsterdam Institute for Advanced Metropolitan Solutions (AMS). Giaccardi: “This particular proto-team is looking at how to make surveillance systems more open and transparent. It’s also looking at how to put together a team to look at all the different levels of interaction that need to be integrated, from algorithms to commercial services and broader issues of political governance.”
PhD candidate Grace Turtle is also part of a proto-team. “For me, the proto-team is a good way to find common interests with other PhD students,” she says. “This creates more connections between the individual projects in a natural way.” Turtle feels very connected to the DCODE network of researchers: “We have summer and winter schools, and informally we communicate a lot through Slack. DCODE is like an adaptive emergent system where researchers group around common topics. I dance between being employed in TU Delft and be a member of the DCODE network.”
Associate Professor of Human-Algorithm Interaction at Delft University of Technology Dave Murray-Rust is trying to integrate DCODE’s philosophy into the master’s program at the Faculty of Industrial Design Engineering, in addition to doing research himself. “For me, DCODE is about educating a new generation of students,” he says. “I hope they change the design world through a more human, happy and critical approach to technology design.”
A DCODE Symposium was held at TU Delft on 22 June, attended by most DCODE researchers. There was also an Interactive Technology Design exhibition where two DCODE prototypes were demonstrated. One was a smart health mirror that allows people to experience a possible future of healthcare in their own smart home. It was an experience for two people. A person, the consumer, stands in front of a mirror that provides information about their health. The consumer can choose how much data he wants to share with the smart mirror and how strictly he wants to follow the health advice.
Another person stands behind the mirror and controls the mirror’s AI system. The AI system provides information about the consumer’s health condition and suggests possible treatments or medications. This other person also has choices, for example, choices that benefit a particular drug company but not necessarily the consumer. Together, the people in front of and behind the mirror can explore scenarios based on different settings of the underlying AI system.
“This prototype is a great way to help people understand how they relate to the algorithmic community,” says Murray-Rust. “Do we assume that the smart mirror only advises us on the right drugs to make our lives better? Or do we recognize that all the interactions we have are seen by drug companies who have their own agenda, which is not necessarily the same like ours?”
Murray-Rust is also preparing a new course for graduate students called ‘Interdisciplinary AI Research Methods’, inspired by DCODE. The course starts in February 2023 and addresses questions such as: can algorithmic impact analysis be a research method for designers to understand AI in practice? Can we use role playing to think about the systems we want to design? Can we look at software and source code and use it to tell stories about system design and development?
Regarding design students’ interest in AI, he says: “A third of the students are very interested in AI. A third are interested but afraid. They don’t know what it is and how to engage with it. And a third actually don’t want to have a lot to do with it.”
Murray-Rust himself started his academic career as a computer scientist, but found over the years that the most interesting questions came from the interaction between computers, technology and people: “The way people relate to technology is becoming more and more fundamental to our society With DCODE we contribute to shaping that future.”
DCODE in a nutshell
DCODE is a European network and PhD programme, funded with four million euros by the EU Horizon 2020 programme. It runs for three years: from 2021-2023. DCODE’s mission is to train researchers and designers to guide the digital transformation of society towards an inclusive and sustainable future. DCODE examines how the use of artificial intelligence in digital products and services is changing design vocabulary and methods, and thus the design practice of the future.
DCODE brings together 40 researchers (including 15 PhD students) from 20 countries. They come from disciplines such as industrial design, engineering, social sciences and humanities. The 15 PhD students work in 8 European cities: Delft, Umeå, Edinburgh, Eindhoven, Aarhus, Copenhagen, Riga and Amsterdam. In addition, DCODE has five laboratories, built around the five grand challenges identified by the consortium, each with a specific focus: the Human Algorithms Laboratory, the Intricate Interactions Laboratory, the Multifaceted Values Laboratory, the Digital Sovereignty Laboratory and the Future Design Practices Laboratory. DCODE also collaborates within various public-public and public-private partnerships.
Summer and winter schools are an important part of DCODE. In June 2022, TU Delft organized a summer school with a special DCODE symposium on 22 June 2022.