2024
Civic AI
Civic AI is a conceptual framework for a responsible AI agent designed to assist citizens in navigating life events. This project explores the intersection of artificial intelligence and civic empowerment, focusing on transparency, control, and user agency in AI interactions.
context
Master thesis @Umeå Institute of Design
team
single project
my responsibilities
Research, Workshop facilitation, Concept, UX/UI, System design
mentorship
I was fortunate to receive invaluable mentoring from Karolina Drobotowicz and Lea Bachmann
[challenge]
Empowering Citizens through Slow AI Interactions
I took on the challenge of designing an AI assistant for public services, drawing from my firsthand experience with their implementation strategies. What sets this challenge apart is its focus on citizen empowerment and agency rather than pure efficiency. It addresses the complexities of major life events and interactions with public services.
Take a minute with me and lets imagine
You got a job offer in a new country—let's say in Denmark. And now, you're planning to move there and it feels super exciting. But at the same time: it's a new country, a new system, and you're suddenly overwhelmed by the sheer amount of bureaucratic things you have to go through.
But thanks to the latest advancements in tech, there is this magic button that promises, you only have to press it and everything is done. So you click it.
Then you go about your life, you're packing your boxes. And suddenly you receive a letter. And that letter states: your visa was declined—and you have no idea why, and no one else does. So, what you realize now is how much power you have given to this button.
And this is a metaphor for what happens to many people in the world who fall victim to decisions made by an AI algorithm. Which is a very disempowering experience. Especially in the public sector, in which we cannot escape the influence such decisions have on our lives.
So, wouldn't you want to have your power back?
[speculation]
What would a civic future look like that is empowering?
A speculation on the future: A Slow AI agent operating within a system co-owned by citizens and the state.
In my thesis I went through a future exploration. And explored what it means to shift the focus from efficiency to agency. From delegating the task, to being empowered to do the task yourself. And with that to recoint the narrative of fast AI, into slow AI.I explored those values by creating an example of a public assistant that guides you through major life events. This life event can be moving to a new country, becoming a parent, or planning your retirement.
It does so by
Firstly, giving you a centralised access to your public information.
Then, engaging in a conversation with you about your new life situation.
Throughout this conversation, transparency is key: the agent clarifies which data points are used and how it knows them. You have control over these data points, including who they're shared with and the option to delete them.
In every interaction, you have the power to negotiate the assistant's behaviors.
With the information shared, it curates relevant content and guides you through the next steps.
If issues arise, it proactively uncovers them. The agent isn't just an additional feature in an existing interface—it's an integral part of the system.
Embedded in a system
Ultimately, the conversation about empowerment led to a system in which Civic AI builds its own trustworthy entity that operates between the people and the state.
Watch the pitch here:
[how I got there]
Process
.jpg)
Research
Conducted expert interviews, literature reviews, and workshops to gain insights into public AI services and citizen empowerment needs.
Concept Development
Employed design sprints and co-creation workshops to craft the Slow AI concept and define key interaction moments.
Prototyping and Testing
Developed speculative scenarios and interaction prototypes, then tested them with citizens and experts to gather feedback.
Refinement
Refined the concept based on feedback, with a focus on enhancing transparency, user control, and agency in AI interactions.
[reflection]
To me, the project was successful as it fostered very fruitful and informed discussions on the topic of civic empowerment in the realm of AI implementation strategies for the public sector.
The biggest challenge was providing concrete solutions in a rather abstract space. However, by working through this puzzle, I gained deep understanding in designing for public services, AI, and value-driven design.
Furthermore, I'm thankful for the shared journey my class and I took this year, helping each other at every turn. I was also fortunate to have great mentors guiding me through the process, as well as insightful stakeholders, interviewees, and participants informing the design process. Thank you to everyone involved!