M
Designing AI Ethics Maturity interactive user interface.
AI Ethics Maturity Model interface, based on research from the University of Rotterdam, was designed to help government entities and organizations in measuring their AI Ethics Maturity level and achieving their goals with greater ease.
Client:
Innovalor
My Role:
Product Designer
Year:
2023
Service Provided:
Product Design, UI, UX
WHAT IS AI ETHICS MATURITY MODEL
The AI Ethics Maturity Model was developed as a framework to guide organizations in evaluating and improving their ethical practices with AI.
This case study explores the design process for creating a user-friendly and visually engaging interactive model that helps decision-makers understand and apply the framework effectively.
THE CHALLENGE
TURNING ETHICS INTO ACTION
Organizations, especially governmental ones, often lack tools to assess how ethically they use AI. The complexity of ethical principles makes it hard for stakeholders to identify gaps or measure progress.
Our high level goals were to:
Design a clear and accessible framework for diverse stakeholders.
Translate abstract ethical principles into visual, actionable and measurable steps.
Encourage continuous improvement through an engaging and easy-to-use tool.
MY ROLE
As part of a collaborative team, I conducted stakeholder and user interviews, defined the problem statement and user persona, designed wireframes and prototypes, performed usability testing, and iterated on feedback.
In addition, I also contributed to client presentations, ensuring the final solution was user-centered and aligned with organizational goals.
THE DISCOVERY
RESEARCH AND ANALYSIS
At the outset of the project, we faced the challenge of not being able to interview government officials directly. Instead, we partnered with researchers from the University of Rotterdam, who were responsible for creating the framework for the AI Ethics Maturity Model.
EARLY INSIGHTS FROM THE FIELD
To understand user needs, we conducted:
Interviews: With University of Rotterdam researchers and AI ethics advisors to gather insights.
Competitor Analysis: To identify industry standards and areas for differentiation.
DEFINING THE PROBLEM
USER PERSONA AND PROBLEM STATEMENT
Our user persona became a Data Analytics Lead in a governmental organization.
DEEPER INSIGHTS
IDENTIFYING OBJECTIVES
Before we could jump into designing, it was important to define
success by setting clear goals which helped us to stay focused on delivering an experience that is intuitive, effective, and aligned with the project’s overall vision.
INFORMATION ARCHITECTURE
Before starting designing it was important to organize the framework into a clear, hierarchical structure.
The subject was very complex and wer had to be sure we fully understood the concept and the framework.
To do so , we broke down the model into:
Maturity Levels (e.g., Foundational, Intermediate, Advanced).
Ethical Pillars (e.g., Transparency, Accountability, Fairness).
Actionable Steps within each pillar to help users progress.
In order to have a clear roadmap of what we will design, we had to understand our user flow. This was quite a difficult process because we had nothing to model the flow on as something similar did not exist.
IDEATION
We used the Crazy 8’s method to quickly brainstorm and generate a variety of design concepts, encouraging creativity and diverse ideas. We selected the most promising concepts through voting and discussion.
A spider-web-style visualization stood out as the best idea, providing a clear and engaging way to show the model's interconnected aspects.
KEY FEATURES
We prioritized features using the MoSCoW method, ensuring a clear focus on what was essential, desirable, and optional for the framework.
THE WHEEL OF ETHICS
One key outcome was the creation of the "Wheel of Ethics," a dynamic visual tool designed to represent both the current and target AI ethics maturity levels.
UNDERSTANDING THE FLOW
DESIGN
FROM SKETCHES TO STRUCTURE
From this point low fidelity wireframes were pretty easy to create. We directly went from designing low fi to mid fi and opted for testing on the mid fidelity wireframes.
MOODBOARD BUILDING
Since we were targeting Dutch institutions we wanted the visuals to represent blend of Dutch traditional imagery and contemporary culture. At the same time we wanted to emphasize the ethical journey within a historical context.
STYLETILE
This approach aimed to create an inclusive design that meets WCAG guidelines and enhances readability for all users.
WIREFRAMING
TESTING AND SHAPING THE VISION
We asked users to try out our design and tell us what they liked and didn’t like. They wanted levels they couldn’t use to disappear and the ones they achieved to stand out in bright colors.
We also found out that our goal-setting tool was confusing. So, we changed it from a bar in the info card to easy-to-use sliders next to the Wheel, making it much clearer for everyone.
Initial Idea: Show out-of-scope levels in dark colors.
User Feedback: Users preferred these levels to disappear and wanted achieved levels to be displayed in stronger colors.
Initial Idea: Change goals within a bar in the info card.
User Feedback: The goal-setting feature was not intuitive; we divided it into slider bars next to the Wheel for better clarity.
PROTOTYPING
BRINGING IDEAS TO LIFE: FINAL TESTING
We tested our high-fidelity prototype to see how users felt about the design and gathered helpful feedback. Users liked how clear and easy to use it was.
We learned that we needed to separate strategic and operational users and make progress on the Wheel easier to understand. To avoid confusion, we added monthly tasks and used tools like spider diagrams to help users track their goals more easily.
THE IMPACT
POSITIVE RESULTS
Our final prototype became more than just a design—it transformed into a powerful tool for introducing and testing the AI Ethics Maturity Model.
With clear timelines and integrated goals, the prototype supports strategic planning and operational alignment. As an educational tool aligned with EU AI ethics laws, it offers organizations a practical pathway to strengthen their ethical practices and compliance.
The results spoke for themselves:
users spent 48% less time trying to understand the model
Engagement during testing sessions increased by 18%
usability scores jumped by 27%, proving the design was intuitive and effective
We measured success by tracking key metrics such as the time users spent understanding the model, the increase in user engagement during testing sessions, and the improvements in usability through direct feedback. These insights helped us refine the prototype and ensure it was both user-friendly and aligned with the intended goals.
