Subsequent to a proliferation of documents to encourage and achieve ethical AI, it becomes increasingly evident that these efforts must be followed by clear actions. However, oftentimes the gap between Ethics Principles and policy is considerable and implementation can be difficult.



One of the key documents for the European Union’s approach towards human-centric and trustworthy AI has been the Ethics Guidelines for Trustworthy AI, presented by the European Commission’s independent High-Level Expert Group on AI. This document has arguably had a comparably stronger impact than many other, similar documents on policy-making, especially within the EU. It therefore lays the groundwork for a new paper investigating ‘Actionable Principles for AI Policy’.



The paper analyses the strengths and weaknesses of the development process of the Ethics Guidelines and draws from that the core elements towards a new framework instigating the development of more implementable ethical principles for the governance of AI. In doing so, it proposes a novel framework for ‘Actionable Principles for AI Policy’ built on three core elements: (1) preliminary landscape assessments; (2) multi-stakeholder participation and cross-sectoral feedback; and, (3) mechanisms to support implementation and operationalizability. These three elements serve to support the real-world applicability and policy implementation of the developed principles by way of placing them within their given environment (societal, geopolitical or otherwise), ensuring that a diversity of voices is included in the process, and, through a concretisation of target audience and suitable mechanisms. 



Inevitably, there are a multitude of other hurdles that AI Ethics Principles face and will face over the coming years. Nevertheless, the pacing-problem is real and future governance efforts will significantly rely on existing (academic) work as they make sense of direly needed policy options. This puts Actionable Principles in a potentially powerful political and societal role, one that needs to be taken seriously and nourished.



Original research paper can be found here:https://link.springer.com/article/10.1007/s11948-020-00277-3 



Stix, C. Actionable Principles for Artificial Intelligence Policy: Three Pathways. Sci Eng Ethics27, 15 (2021). https://doi.org/10.1007/s11948-020-00277-3