Why the Hardest Moral Problems are the Easiest to Solve: What Ethical Ai tells us about Moral Life
The Social Theory and Artificial Intelligence: Challenges, Opportunities and Futures Seminar Series from the BSA Theory Study Group
1 March 2027 (13:00-14:00 GMT)
Online
About the Event
Speaker
- Dr Owen Abbott, Lecturer in Social Sciences, Cardiff University
AbstractWhat do attempts to develop 'ethical Ai' (EAi) systems tell us about morality? And what do social theories of morality tell us about issues in how EAi models are developed?
Generations of moral philosophy has agonised over questions of what the optimal form of moral reasoning is in certain situations and how life-and-death moral dilemmas should be judged. Now, in our new Ai age, models of EAi have developed complex systems of optimised moral reasoning and moral judgements based on large quantifiable data (Jiao et al., 2025; Hendrycks et al. 2023).
Famously, the Moral Machine Experiment gathered global data on what moral decisions autonomous vehicles should make by collecting millions of answers to life-or-death “trolley problem” moral judgement experiments (Awad et al. 2018). In some respects, such work has used big data and Ai modelling to provide “answers” to the “hardest” moral problems. However, EAi models (including ordinary-level EAi systems like Operation Delphi) are confounded by the kind of everyday moral situations that require contextual interpretation, which humans deal with routinely. This is because EAi has largely been developed according to computational models that prioritise data-driven approaches and facilitate optimisation and maximizing preferences (Tasioulas, 2022).
I argue that EAi development illuminates the limitations of quantifications of morality and reveals how the complexity of human moral interpretation and practice means everyday level moral problems are less amenable to being “solved” than the “hardest” moral problems. I thus show that practice-based social theories of morality are needed to reframe EAi towards human capacities for interpreting nuanced moral situations.
Further Information
There will be a 30 minute presentation followed by a 30 minute Q&A session. Due to BSA’s policy, this event will not be recorded.
Registration
This event is free to attend but registration is required.
Contact Us
Please contact Dr Zheng Liu with any queries.