AGENT-BASED MODELLING FOR PUBLIC SOCIAL SERVICE DISTRIBUTION

Petra Ahrweiler, Nigel Gilbert, Martha Bicket, Albert Sabater Coll, Elisabeth Spaeth, Hassan Bashiri, Ebin Deni Raj, Blanca Luque Capellas

Keywords: Social Assessment, Public Service Provision, Context-Specific ABM

TL;DR: Using context-specific ABM for social assessment in public social service distribution

Abstract: The paper introduces an agent-based modelling approach for assessing beneficiaries in public social service distribution. The {Anon} project combines empirical research, gamification, and agent-based models (ABM) to assess the fairness of AI-based distribution in different countries and propose improvements. The paper presents a participatory research strategy, where ABM and serious games are used to identify more desirable social assessment routines within heterogeneous cultural contexts. By following this approach, the paper suggests that context-specific ABM can be used for co-designing AI systems with stakeholders, to assist ex-ante evaluation for testing and prototyping AI systems before implementation thus reducing risks and costs, and for scenario analysis and asking what-if questions to reduce uncertainty. Besides offering a useful tool to help social workers in reflecting on and improving their assessments in their immediate workplace, the proposed ABM-centred approach is also relevant more generally for public, social and technology policy. Prototyping helps to avoid risk of failure, unintended consequences, and systems that turn out to be ineffective following expensive development. The option to address what-if questions, to test interventions before implementing them, and to evaluate the advantages and disadvantages of different scenarios, is of great relevance in many policy domains.

Submission Number: 4