PhD Topic: Open Models for AI/Data Risk Assessment (DC5)
Research objectives: The candidate will (i) investigate open models to discover, represent and mitigate ethical and legal risks raised by the EU’s AI Regulation and Digital Services Act; (ii) capture the information flows required for AI certification and market surveillance between AI providers, AI users, AI importers and regulators and stakeholder’s whose health, safetyor fundamental rights may be impacted; (iii) develop an open semantic model to exchange information on risks assessments, risk treatments and risk impacts information in support of multi-stakeholder learning in specificverticals (iv) develop and assess a tool to capture, exchange and analyse risks for a specific high-risk verticals to support multi-stakeholder agreement on e.g. probabilistic risk thresholds and reasonably foreseeable forms of risk/misuse.
Expected Results: (i) a machine-readable ontology for risk and impact information exchange; (ii) design of methods and tools to extend the ontology in support of multi-stakeholder regulatory sandboxes for high-risk AI areas
The candidate is expected to make a research stay at Trinity College Dublin.