A norm optimisation approach to SDGs: tackling poverty by acting on discrimination

Georgina Curto, Nieves Montes, Carles Sierra, Nardine Osman, Flavio ComimIJCAI 2022 · DOI

TL;DR

This paper proposes a novel approach to poverty reduction by tackling "aporophobia" – the discrimination against the poor. Using agent-based social simulations, they aim to understand how norms of equal opportunity and social solidarity interact with aporophobia and influence the effectiveness of poverty reduction policies. The ultimate goal is to develop guidelines and an "aporophobia-meter" to inform future policy-making.

What Problem Does This Paper Address?

Traditional poverty reduction policies, while aiming for distributive justice through equal opportunity and solidarity, have proven insufficient. A significant issue is the rise of "aporophobia," a prejudice against the poor, which can stem from the rhetoric of meritocracy in an era of increasing inequality. When individuals are blamed for their lack of progress, it can lead to resentment and hinder the effectiveness of poverty alleviation efforts. This paper addresses the gap in understanding how aporophobia acts as a barrier to poverty reduction and how to design policies that mitigate it, aligning with the UN's "Leave No One Behind" principle.

Methods

The core of this research lies in developing an agent-based social simulation model. This model will incorporate:

  • Aporophobia Conceptual Framework: Building on existing literature and conducting a new survey to empirically understand aporophobia.
  • Social Simulation: Agents will interact within a simulated society, with their behavior influenced by norms related to equal opportunity and social solidarity, as well as their individual level of aporophobia.
  • Meta-heuristic Optimization: Techniques like Genetic Algorithms (GAs) will be used to search for optimal levels of prescriptive norms (policies) that minimize poverty and aporophobia, aligning with defined values.
  • Value Semantics: Defining how abstract concepts like "minimum poverty" and "maximum fairness" (non-aporophobia) are operationalized within the simulation.

The methodology builds upon previous work on value-guided synthesis of parametric normative systems, with a key innovation being the explicit inclusion of aporophobia as both a macro-level indicator and a micro-level agent attitude.

Datasets

The paper mentions the intention to conduct a new survey to gather empirical data on aporophobia. The raw data from this survey will be made publicly available, anonymized, through a dataset platform. No other specific datasets are mentioned for training or evaluation.

Key Insights and Ideas

  • Aporophobia as a Policy Barrier: The central idea is that aporophobia, not just lack of resources, significantly hinders poverty reduction efforts. Blaming the poor for their situation can undermine well-intentioned policies.
  • Norms and Aporophobia Interaction: The research will explore how different combinations of "equal opportunity" and "social solidarity" norms affect aporophobia levels and, consequently, poverty reduction outcomes.
  • Beyond Resource Redistribution: The project moves beyond traditional economic approaches to poverty by incorporating psychological and ethical dimensions, specifically the social stigma associated with poverty.
  • The "Aporophobia-Meter": A novel SDG indicator is proposed to measure and track the evolution of aporophobia, providing a tangible tool for policymakers.
  • Focus on the "Non-Poor": The approach recognizes that addressing aporophobia requires engaging not only the poor but also the broader society.

Key Results

Not discussed in the paper, as this is a proposal for future research. The paper outlines the intended methodology and expected outputs, including a conceptual framework, simulation results, and policy guidelines.

Limitations and Future Work

  • Ungrounded Social Model: The authors plan to mitigate this by grounding the model in state-of-the-art literature and their own empirical survey data. They also emphasize adapting the model to local characteristics to avoid bias from a single demographic.
  • Biased Social Model: To address this, they plan to conduct the survey globally and tailor recommendations to local contexts, paying special attention to historically discriminated groups.
  • Irrelevant Results: A policy review of regional and global frameworks is planned to ensure the guidelines are relevant to international NGOs and government officials.
  • Lack of Financial Resources: The team has incorporated the project as an internal objective and applied for external funding.
  • Lack of Computing Power: They plan to leverage existing HPC services and implement parallelizable search strategies like Genetic Algorithms.
  • Defining Value Semantics: Acknowledged as a challenging task, the definition of "minimum poverty" and "maximum fairness" will be crucial and will be informed by literature and potentially new indicators.
  • Agent Aporophobia Modeling: The computational construct for modeling aporophobia at the micro-level will be refined after the conceptual framework and empirical survey are finalized.
  • Human Acceptance of Norms: Future work will involve evaluating whether humans accept the derived norms and if exposure to the research reduces their aporophobia levels.

Summary updated

BibTeX
@inproceedings{curto2022norm,
  author = {Georgina Curto and Nieves Montes and Carles Sierra and Nardine Osman and Flavio Comim},
  title = {{A norm optimisation approach to SDGs: tackling poverty by acting on discrimination}},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year = {2022},
  doi = {10.24963/ijcai.2022/726},
}
Figure