Artificial Intelligence-Based Welfare Distribution and Universal Basic Income Readiness: A Comparative Analysis of Digital Subsidy Systems in Emerging Economies

Abstract

In emerging economies, digital public infrastructure now plays a pivotal role in the delivery of welfare programmes, especially when governments want to integrate the delivery of digital identity, of payment systems, of social registries, of mobile connectivity and of the targeting with data into a scalable welfare system. There is growing interest in using AI as a means of identifying beneficiaries, detecting anomalies, preventing fraud, monitoring payments and triaging grievances, and in tackling exclusion errors, subsidiarity issues and labour market uncertainty through the implementation of UBI. This paper looks at the link between ready-to-use UBI and AI-driven welfare distribution by reviewing digital welfare distribution systems in India, Brazil, Indonesia, South Africa and Kenya. It has a comparative quantitative design with a synthetic data set of 1000 simulated welfare beneficiaries in each country with 200 observations for each country of interest. Indicators are included in the model for 1) access to digital ID, 2) financial inclusion, 3) mobile internet access, 4) digital literacy, 5) payment delay, 6) accuracy of targeting by AI, 7) risk of exclusion, 8) possibilities on grievance redressal, 9) beneficiary satisfaction, and 10) preparedness for UBI. The simulation results indicate that the welfare effectiveness is positively correlated with AI-targeting accuracy, and exclusion risk is negatively correlated. The one-way ANOVA shows differences in welfare effectiveness across countries with F (4, 995) = 29.22, p < .001. The findings from regression analysis indicate that targeting with AI and digital literacy, in addition to grievance redressal, are facilitating the effectiveness of welfare agencies, whilst the role of digital identity alone, if a digital identity exists, is insignificant without accessible avenues for correction and appeal. The paper offers a clear simulation model for digital welfare care systems, which strives to compare care systems without (falsely) presenting simulated evidence as primary evidence. It represents a mixed approach of more narrowly targeted forms of subsidies, alongside some elements of a basic income and with governance principles of AI aligned to human ends, (ict uses inclusion, fiscal realism, and accountability).