The digital has become an intrinsic part of our lives — influencing and tying to every aspect. Many times, it is the first stop for vulnerable populations (e.g. survivors of domestic violence) to seek support services. Organizations are slowly turning to AI to reach vulnerable populations where they are, i.e. online. For instance, the Sis Bot chatbot in Thailand — which can be accessed both through phones and computers — provides survivors of violence access to information around the clock. Similarly, in Mongolia, chatbots are being integrated into Facebook messenger, making them more accessible.
The COVID-19 pandemic is considered a ‘shadow pandemic’ for Gender-Based Violence (GBV). Alarming data across countries showed an increase in calls to domestic violence helplines in several countries during this time. With stringent lockdowns, access to support services and information became difficult. The COVID-19 pandemic saw an increase in Internet usage by 50% to 60%.
In light of this, Quilt.AI explored the needs and requirements of underrepresented, vulnerable populations using its vast AI capabilities. Our research on gender — specifically violence against women and girls (VAWG) — draws attention to the scope of online data of this nature in reaching survivors online. We studied VAWG in eight countries in South and Southeast Asia: Thailand, Malaysia, Indonesia, the Philippines, Sri Lanka, Nepal, India, and Pakistan. Quilt.AI’s Cultural AI tool analyzed hundreds of social media posts and search keywords.
Here are our key takeaways based on our AI analysis of VAWG during the COVID-19 pandemic (the full report can be accessed here):
Online behavior demonstrated by survivors can be used to identify their key needs. During periods of crisis with mobility restrictions, online platforms can be useful for research and implementation. Our Knowledge-Awareness-Intent behavior change framework uses search behavior (gauged through search engines such as Google) to identify people’s unspoken needs. It can also be used to triangulate data from on-ground sources. For instance, our analysis showed an increase in search queries as search queries related to VAWG, including physical, sexual and psychological violence, during the study period. These include keywords such as: “physical abuse signs”, “violent relationships”, and “cover bruises on face”.
Not only this, it brought to light the different kinds of search trends across various countries. Here are some examples of keyword searches by country:
Local language searches helped cast a wider net and understand the pulse of the larger population, which may be more familiar with languages and dialects other than English.
Our analysis extended to more vulnerable populations affected by stringent lockdowns, i.e. migrant workers. Many of them were stranded in locations far from home and sought support from their peers through safe online spaces. Insights from the social media analysis spotlighted support groups on platforms such as Facebook. Indonesian migrant support groups (for Indonesian overseas workers) were identified along with Youtube posts by migrant influencers. These also become spaces where CSOs and other organizations can provide pandemic and help-seeking-related informwwation. Similarly, migrants from Myanmar — many of whom lived in Thailand — used Facebook to show solidarity among peers and to stay informed about pandemic-related rules and case numbers in their countries of work. Identification of these women’s workers and peer groups can be helpful to CSOs for disseminating important information and reaching people in geographically far-off locations.
Technology, specifically AI, offers tremendous potential to organizations regarding crisis intervention and response. The Internet should be viewed as a crucial repository of valuable data that can be harnessed to address the world’s most pressing problems.
The SOAS Centre for AI Futures, in partnership with SOAS, Quilt.AI, and the Helsinki Institute of Social Sciences and Humanities, has developed AI-powered research capabilities to denoise millions of Internet data points into insights for social good. Learn more here.
.svg)
