At the highest level, our approach is characterised by two main ideas

  • We focus on understanding governance throughout the design and deployment process? In particular, this calls for us to address the complete lifecycle of iterative and multi-actor AI development
  • We focus on understanding how and why systems fail (interpreting the term broadly), and on developing tools to gather evidence for regulators/developers

Application Domains

The framework we develop is intended to be broadly applicable, across numerous autonomous systems domains. Our activities will aim at demonstration with case studies in the following domains:

  • Mobility, e.g., Beyond Visual Line of Sight Operation of UAVs
  • AI in medical domains, e.g., screening and diagnostics decision making
  • Automated systems in the social care sector

Structure of the Node’s Activities

Our work is organised modularly so as to strike a balance between the breadth of expertise and concerns, and the need for integration and cross-pollination. This includes:

  • Multiple pillars: (1) legal and social studies; (2) qualitative and quantitative studies of responsibility, and causal modelling (3) Computational tools using formal methods, NLP, etc. and (4) design ethnography as a cross-cutting theme
  • Two Phases (24 months and 18 months respectively): First develop framework and tools with internal testbeds, then work with external partners and regulators to deploy/certify full cycle case studies