Evaluating AI-Enabled deception vulnerability amongst Sub-Saharan-Africa migrants

TL;DR

This study assesses how Sub-Saharan African migrants' ability to distinguish AI content influences their vulnerability to AI-enabled scams, highlighting prior targeting exposure and verification effort as key factors.

Contribution

It introduces a hybrid modeling approach to evaluate AI literacy's impact on deception vulnerability among SSA migrants across different transnational contexts.

Findings

Prior exposure to targeting increases scam vulnerability

High verification effort reduces vulnerability

Transnational factors like duration abroad have minimal impact

Abstract

In this study, the vulnerability of Sub-Saharan African migrants to AI-enabled deception, specifically the risk of exposure to scams targeting them, was evaluated. I hypothesized that the ability to distinguish human-generated content from AI-generated content had far-reaching implications beyond content assessment to determining vulnerability to AI-enabled deception. Data collected from a survey of 31 professionals and migrants from SSA across Europe and North America, covering themes on Demographics and Transnational Context, Core AI Literacy and Vulnerability, Mitigation and Trust, was modelled using a hybrid Structural Equation Model and Multiple Linear Regression. The results indicated that the strongest indicator of vulnerability to AI-enabled deception, such as scam, was prior exposure to targeting, as targeting has previously been noted to be, in most cases, a calculated attempt. Confidence in the ability to identify AI content as well as the behavioral characteristics of high verification effort, emerged as significant protective factors that could lower the vulnerability to AI enabled deception. Other transnational contexts such as duration spent abroad or engaging in international fund remittance were found to have a small and insignificant effect on vulnerability.