AMS institute aims to research, design and test technology-driven solutions to deeply understand and solve the metropolitan challenges of the city of Amsterdam. They aim to address big urban challenges in the fields of energy, circularity, digitization, climate resilience, and mobility.

As part of a volunteering collaboration with the TU Delft, we were approached to conceptualize an AI-driven solution for reducing the city’s energy consumption. We specifically were tasked to envision a system that aligned automated decision-making with the values of transparency and contestability, ensuring responsible design with AI.

We created a Smart City Lights system that uses machine learning to reduce energy consumption in the urban environment of Amsterdam. Through contestability by design, we analyze and prevent possible harms, and give stakeholders the right to influence automated decisions that impact them directly or indirectly. We also drew attention to the complex dynamics between such a system and the wider societal context to further ingrain practices that prevent harmful system behaviour.

You can read the full paper here.