About the project
This project will develop computational methods to detect responsibility gaps, attribute collective failures, and support safer oversight of multiagent systems under uncertainty and partial observability.
AI safety becomes harder when decisions are distributed across autonomous agents controlled by different people or organisations. These agents may be software services, robots, decision-support systems, or human decision makers. Harm can emerge even when no single agent clearly causes it, especially when knowledge is incomplete, tasks are delegated, and several agents could jointly intervene. Existing approaches often analyse causal contribution, assigned duties, or strategic ability separately, leaving gaps between who caused an outcome, who could have prevented it, and who ought to answer for it.
This project will develop an integrated framework for detecting and addressing responsibility gaps in multiagent systems. It will examine three linked challenges:
- responsibility for unanticipated harm and negligence
- collective responsibility when intervention requires a coalition
- attribution under uncertainty and partial observability
You may focus on one or more challenges while contributing to a shared formal model. Methods may include game theory, temporal, epistemic and deontic logics, formal verification, theorem proving, automated reasoning, and agent-based simulation. The work will connect prospective responsibility, identifying who must monitor or intervene before harm, with retrospective responsibility, attributing failures afterwards.
Expected outputs include formal definitions, attribution algorithms, an automated reasoning prototype, formal guarantees, simulation-based evaluation, and guidance on the information and logging needed for accountable oversight and control. Case studies may involve AI-assisted healthcare, shared-control autonomous systems, and AI-supported building management.
You will participate in doctoral training and research seminars within Electronics and Computer Science and the Agents, Interaction and Complexity group. Further development opportunities will include academic writing, responsible research and innovation, conference participation, public engagement, and interaction with researchers and stakeholders working on responsible AI and autonomous systems.
The School of Electronics and Computer Science is committed to promoting equality, diversity inclusivity as demonstrated by our Athena SWAN award. We welcome all applicants regardless of their gender, ethnicity, disability, sexual orientation or age, and will give full consideration to applicants seeking flexible working patterns and those who have taken a career break. The University has a generous maternity policy, onsite childcare facilities, and offers a range of benefits to help ensure employees’ well-being and work-life balance. The University of Southampton is committed to sustainability and has been awarded the Platinum EcoAward.
