IGF 2022 WS #425 Data divide worldwide: addressing data scarcity and bias

Description:

With 2.9 billion people still unconnected to the Internet – those in least developed countries, rural communities and vulnerable groups, in general, being the most affected –, multiple challenges can be identified: not only do these people lack an access that has become a prerequisite for participating in the contemporary society, education and economy, but also they are underrepresented in the datasets used to support decision-making at different levels, as well as for machine learning, thus increasing algorithm bias. A striking example of this is the lack of datasets that represent people with disabilities, limiting the progress in developing IA solutions for these populations, or at the very least, not amplifying inequalities via biased algorithms. On top of this, another layer of data scarcity is that of statistical data. According to OECD, no data exists for approximately two thirds of SDG indicators. The lack of timely and relevant data on ICT access and usage mostly affects countries in the Global South and the most disadvantaged populations within them, making it more difficult to accurately portrait their situation or to monitor policies affecting these populations. These multiple layers of inequalities can be conceived of as data divide.
When communities are underrepresented in the data, decisions made based on this data may both overlook their unique needs and/or lead to skewed conclusions. Nowadays, both structured data (mainly statistical data) and non-structured data (mainly user-generated data, or big data) are recognized to be complementary and necessary for policymaking and for tracking the SDG. However, the timely and meaningful access to both sources of data involves specific challenges. For big data, unequal access to broadband service, variations in access to different types of devices, disparities in digital literacy can influence who is included in the data and who is not. Moreover, information produced in a particular area of a given country can serve as an indirect indicator of the extent to which that community can capture the benefits of the data revolution. In the case of structured or statistical data, challenges range from budgetary to technical capacity and political will.
For strengthening a measurement ecosystem and the meaningful integration of different sources of data into policymaking, on the one hand, the full scope of the issue needs further conceptualization, research and understanding and, on the other hand, multistakeholder agreements for data access, and data governance mechanisms still need to be reached. This workshop will address these issues, explore novel notions such as 'datashpere', 'data divides' and other concepts through which an understanding of these new phenomena can be achieved. Reflections on their social impact and the way forward to bridge these divides will be shared from the perspectives of stakeholders from Africa, Europe, Latin America and North America.