What Are the Limits of Predictive Algorithms in Public Policy?
Academics
- Julie George, Institute of Health Informatics, Population Health Sciences
- Sofia Olhede, Statistics/Big Data Institute
- Dr Henry Potts, Institute of Health Informatics
Algorithms derived from big data are increasingly being used to inform resource allocation decisions across the public sector. But can we ensure that machine-generated decisions are fair towards different population groups? If not, then their use might widen social inequalities or breach equality laws.
This interdisciplinary project examined the ethical and practical boundaries of using predictive algorithms in public policy. A half-day workshop (6 June 2018), funded by UCL Grand Challenges, explored the technical, ethical and legal issues around the use of predictive algorithms in resource allocation. This well-attended workshop that brought together a range of academic and professional perspectives on algorithmic fairness. The event led to new research directions and collaborations, a workshop report and plans for a joint academic publication.
Half Day Workshop
Algorithms and Fairness: What are the limits of predictive algorithms in public policy?
Wednesday 6th June 2018, 13.00 – 17.00
Speakers
- Setting Out the Problem Julie George (Institute of Health Informatics) & Sofia Olhede (Big Data Institute)
- Priority Setting And Predictive Algorithms: A Match Made in Heaven? Benedict Rumbold (Department of Philosophy, UCL)
- Machine Learning and Social Learning Jack Stilgoe (Department of Science and Technology Studies)
- Distributional Cost-Effectiveness Assessment Susan Griffin (Centre for Health Economics, University of York)
- Casual Assumptions in Fairness Ricardo Silva (Department of Statistical Science, UCL)
- Algorithms, Equity and Law Sophia Adams-Bhatti (The Law Society)
- Reflections on Research Agenda for the Public Sector Indra Joshi, Digital Health & AI Clinical Lead, NHS-England
Group Discussions
- Developing a Research Agenda
- Next Steps
Next Steps:
- Publish the workshop report
- Submit a joint peer-reviewed paper
- Apply for funding to explore statistical modelling in medical recruitment fairness
Outputs and Impacts
- Academic Collaboration: Strengthened cross-departmental links and initiated new research partnerships.
- Teaching Innovation: Workshop insights contributed to teaching on AI in Public Health within an MSc module.
- Societal Engagement:
- Informed a local authority funding bid on ethical use of natural language processing in social care
- Contributed to NHS discussions on AI ethics
- Engaged with BMJ staff on algorithmic fairness in publishing
- Disseminated via social media and informal public engagement
Close
