Code 8.7: Vulnerability Mapping and Modelling
The Code 8.7 showcases, panels, plenary sessions and mini-hothouses had a number of common themes in addressing how and in what ways computational science and artificial intelligence (AI) can help the fight against modern slavery. Across the two days of the conference, there was much discussion about the hidden nature of the phenomenon of modern slavery and from a scientific perspective, the fundamental problem of “unobservability”. Slavery is akin to other human rights violations in that agents and perpetrators are differently motivated, victims and survivors are hard to find and acts of commission and omission that lead to increased vulnerability and actual instances of modern slavery are difficult to document and analyse.
From left to right: Harry Cook (IOM), María Olave (Iniciativa Regional América Latina y el Caribe Contra el Trabajo Infantil), Davina Durgana (Walk Free Foundation), James Goulding (Rights Lab).
Computational science and AI offer partial solutions to these problems by using direct and indirect observations that are known about modern slavery to provide inferences on and insights into activities, people and trends that are as yet unknown. Like other social scientific research, such an evidence-inference methodological core sits at the heart of any solution offered by the myriad of ways computational science and AI can provide additional explanation and understanding to the problem, including mapping vulnerability to slavery, prevalence of slavery, roots causes of slavery, liberation of victims and prosecution of perpetrators.
Across the presentations and discussions there was an underlying need for clarity about how anti-slavery organizations, government agencies, academics and technology companies think about operationalizing modern slavery for careful and systematic analysis. One useful framework comes from Adcock and Collier, the elements of which help address many of the problems and issues raised at the conference, and which has been used previously in work on measuring human rights. The framework has four main elements:
- background concept;
- systematized concept;
- indicators; and
- scores on units.
Background concept: Modern slavery is a concept that is not uncontested, but one for which there is an emerging consensus that has mobilized academics, government agencies, anti-slavery organizations as well as technology and other private-sector companies, who were all represented at Code 8.7. Moreover, Target 8.7 of the Sustainable Development Goals itself is evidence of the global recognition of the problem and commitment to find effective ways to end slavery by 2030.
Systematized concept: The main meaning and core content of what constitutes modern slavery, including its different attributes and dimensions needs to be systematized using international law, international human rights law and the Bellagio-Harvard Guidelines on the Legal Parameters of Slavery. More work is required on using these legal frameworks to delineate a set of core attributes of modern slavery.
Indicators: The attributes and main dimensions of slavery can be operationalized using direct and indirect measures. Direct measures include acts, decisions, violations, cases, victims and perpetrators involved in modern slavery. Indirect measures include perceptions, feelings, physical objects (e.g. brick kilns), behaviours, financial transactions, community practices and cultures and larger sets of structural factors related to the increased vulnerability and probability of individuals falling into conditions of slavery (see also the “determinants of vulnerability” model from the International Organization for Migration).
Scores on units: The actual data generated through application of the indicators to observations across different units of analysis included:
- Events-based data, such as the 91,000 case management data presented by Harry Cook from IOM;
- Standards-based data, such as the Cingranelli and Richards country-level data on the protection of worker rights and the Government Response Data from the Walk Free Foundation;
- Survey-based data, such as individual level data on child labour in Latin America presented by Maria Salvo from the ILO and the Global Slavery Index from the Walk Free Foundation;
- Socio-economic and administrative statistics; and
- Big data, such as the work on vulnerability mapping in Tanzania presented by James Goulding from the Rights Lab and the geospatial analysis of brick kilns using satellite imagery presented by Doreen Boyd from the Rights Lab.
Computational science and AI are predicated on strong theoretical and conceptual foundations that are then used to make initial observations necessary for training and developing algorithms for processing, predictive analytics and inferential statistics applied to large-scale structured and unstructured data. Combining such forms of analysis across different data streams and data types allows for more complete and robust pictures of the direct and indirect indicators of modern slavery.
The work on Tanzania does not measure slavery per se, but it does provide insights into the geographical probabilities of slavery based on varying degrees of vulnerability to slavery. The work on the brick kilns in South Asia does not measure actual incidents of slavery from space, but instead estimates the total number of physical sites where there is a high probability of slavery occurring.
These two examples make the larger point, which was reinforced across other sessions at Code 8.7, that computational science and AI alone are not sufficient for combatting slavery, but that the insights that these methods of analysis provide reveal many of the unobserved and hidden elements of modern slavery that then assist advocates, prosecutors, NGOs and other stakeholders to design and implement purposeful interventions to reduce and/or eliminate enslavement.
Todd Landman is Professor of Political Science and Pro Vice Chancellor of the Faculty of Social Sciences at the University of Nottingham.
This article has been prepared by Todd Landman as a contributor to Delta 8.7. As provided for in the Terms and Conditions of Use of Delta 8.7, the opinions expressed in this article are those of the author and do not necessarily reflect those of UNU or its partners.