Implementing AI Ethics: Making Sense of the Ethical Requirements

Ethics implementation remains largely siloed in technical teams: The study found that middle and higher-level management executives rarely engage with ethical requirements beyond viewing privacy and data governance as legal compliance issues, suggesting that responsibility for AI ethics is not being distributed across organizational leadership where strategic decisions are made.

Most organizations only recognize ethics as legal obligation, not values-driven practice: Among the ten Finnish software engineering executives interviewed, ethical requirements were primarily understood through a compliance lens (particularly EU regulations) rather than as core organizational values, indicating a gap between ethical frameworks like the EU's Trustworthy AI guidelines and how companies actually operationalize them.

Technical teams are translating ethics into familiar business language: The research found that organizations successfully implement ethical requirements when they reframe them as "risk requirements" (for technical robustness and safety) and "sustainability requirements" (for societal and environmental well-being)—suggesting that connecting ethics to existing management frameworks makes implementation more practical.

Management practices need restructuring to embed ethics: The study reveals that current Agile portfolio management frameworks don't naturally accommodate ethical considerations, indicating that organizations need to deliberately redesign their management processes and decision-making structures to make ethics a routine part of how they plan and prioritize AI development.

There's a significant gap between ethical guidelines and management practice: While comprehensive ethical frameworks exist (like EU Trustworthy AI guidelines), the research shows executives lack practical tools and processes to translate these principles into day-to-day management decisions and organizational practices.