As homes begin to lean into autonomous systems to decide when a battery charges or when to trade energy with the grid, households can see what their system did, but not why. This dissertation is a Research through Design based study aiming to explore that gap: what does it actually take for a non-technical participant to understand, trust, and stay meaningfully in control of decisions an AI is making on their behalf?

Working across research, design and engineering, I built a spectrum of four working prototypes, from a plain facts dashboard through to one that justifies its choices via alternative reasoning, and tested them against twelve non-technical participants through think-aloud sessions and interviews.