RAAIT researcher Sietske Tacoma from the AI research group at Utrecht University of Applied Sciences has developed a clear and accessible guide for organisations working with foundation models in AI. The guide, titled Building on Foundation Models, provides practical tools for making informed and responsible decisions when developing AI applications.

The guide consists of a visual “discussion sheet” and an explanatory document that outlines key considerations – including ethical, technical and organisational aspects – when deploying foundation models in real-world settings.

A central question that emerged during the project was: how can end-users retain control over AI outputs? For example, some generated images might include implausible or misleading content, such as a campfire burning inside a tent. This challenge will form the basis of upcoming follow-up research, exploring technical and ethical strategies to better guide AI-generated outcomes.

RAAIT (Responsible Applied Artificial Intelligence) is a collaborative programme involving the Universities of Applied Sciences in Amsterdam, Utrecht, and Rotterdam. Together with partners from the field, RAAIT develops responsible AI solutions that aim to positively impact people, society and the planet.