Ethics as infrastructure
Ethical deliberation is also a productivity tool: it reduces avoidable rework by surfacing constraints early, and it speeds up everyday decisions by giving teams a shared rulebook for “can we do this?” It also lowers governance overhead by making choices easier to justify, document and revisit.
We began by grounding the work in Digivisio’s own value base. First, we reviewed existing materials and commitments: how values are defined, and how those values already show up in goal-setting and practice. We then expanded that foundation by introducing values and perspectives from the field of technology ethics and responsible AI best practice, alongside emerging regulatory requirements for trustworthy AI in the EU. In particular, the EU AI Act, together with related data protection and digital regulation, served as an important reference point. It was a checklist to be complied with, but also a signal of where expectations of responsibility, transparency and accountability are concretely moving. These frameworks bring into focus concerns that are often deferred until systems are already in use, at which point accountability becomes urgent and corrective action far more costly.
From there, the work moved from values on paper to values under pressure. Together with Digivisio’s teams, we unpacked what responsibility means specifically in the context of data and AI use: what needs protection, what needs visibility, what needs boundaries, and where trade-offs are most likely to surface – not by pushing these principles into immediate action, but by giving space to critique, evaluation, and iteration until the principles could land with meaning in their own contexts. Also, the network of Finnish higher education institutions was invited to comment on and contribute to these principles through a feedback process.
In parallel, we engaged Digivisio’s target groups through qualitative interviews with students and other learner communities for a grounded ethics approach to articulate the conditions under which data- and AI-enabled services can be defined as trustworthy, legitimate and worth engaging with. A central outcome of this work was the identification of two key design drivers for responsible data- and AI-enabled learning services:
