To test whether an Autonomous System (AS) is trusted, researchers typically develop a use case – often a one-off never to be used again.
This project will document and critically review scenarios previously used to evaluate human trust in AS and those that could be used in the future (e.g., adapted from human-human trust experiments and as envisioned by TAS artists-in-residence). We formulate a taxonomy and will develop an accompanying online library from which researchers can reference use cases – by domain, and by whatever aspect of trust the use case tests is for (e.g., human willingness to comply, conformance to expectation, etc.).
The TAS Benchmarks Library and Critical Review project is led by Dr Peta Masters, Kings College London and Dr Alan Chamberlain, University of Nottingham.
Project kick off video