Digitalization of social protection has been an emerging trend that is rapidly transforming the way social welfare is delivered to individuals and communities worldwide. The development and use of digital technologies continue to grow and   influence how governments operate and function today. At times, this has led to marked improvements for service users and providers, but the potential downsides have also frequently been evident. This transformation touches on vital elements of the delivery of public services and promotes progress toward numerous Sustainable Development Goals (SDG). In this blog post, we will explore potential solutions to the social and ethical dilemmas related to harnessing information from digital sources. These dilemmas can be divided into three groups: 1) data protection, 2) data poverty, and 3) bias and discrimination. 

Data Protection

Protecting data is crucial for social protection, as it often contains sensitive information that needs to be safeguarded against data breaches and potential cyber threats. Compliance with privacy regulations specific to data origin and storage is essential. Obtaining individuals' consent before data collection and analysis, as facilitated by frameworks such as GDPR, ensures people’s privacy is respected. Individuals should have the choice to opt out of data storage and analysis while still receiving benefits. Employing techniques like pseudonymization and encryption can enhance anonymity. For instance, Facebook's use of limited and aggregated data for aiding food distribution in Chile demonstrates privacy preservation.

Moreover, when utilizing digital data in social protection systems, concerns arise regarding excessive government surveillance and potential harm to certain groups. One example from the United States is the unconstitutional surveillance programme of the National Security Agency’s (NSA)  called PRISM- through which the NSA, the Federal Bureau of Investigation (FBI), and the Central Intelligence Agency (CIA) collected and examined the international emails, internet calls, and chats of Americans without securing a warrant. Mitigating risks involves collecting only necessary data, minimizing data retention, and involving stakeholders and vulnerable community members in programme design. This balances data collection benefits with the risks of increased surveillance. Furthermore, accessing digital data for social protection is challenging due to ownership complexities. Private companies often own data sources, making access expensive or contingent on special agreements. Government data is fragmented across departments, necessitating negotiations with politicians. Collaboration with private companies has been attempted before in order capitalize on efficiency but it can be precarious, granting them substantial influence over assistance efforts. To avoid such problems, emphasis should be placed on universally accessible or organization-owned data sources. The success of this approach has been proved by the case of the free and open Landsat data policy which has the potential to empower social protection programmes by providing access to high-quality satellite imagery for informed decision-making, disaster response, environmental monitoring, and targeted interventions.

Data Poverty

The digital divide poses challenges for social protection systems reliant on digital data sources, as individuals without access to technology such as socioeconomically disadvantaged persons, older people and minority or marginalized communities, are excluded from data collection. As of January 2022, based on DataReportal's Digital 2022 report, approximately 58 per cent of Kenyan residents lacked access to reliable internet. During the COVID-19 pandemic, when the statistics were grimmer, it meant that those individuals had limited access to information and to various government services which transitioned to online platforms. Further, The Mobile Gender Gap Report for 2023 reveals that the percentage of Kenyan men utilizing mobile internet has remained stagnant at 59 per cent, compared to the previous year. Meanwhile, the adoption rate for women has slightly increased, rising from 36 per cent to 39 per cent.

In general, we see a positive trend where more women in low- and middle-income countries are embracing mobile internet. However, it's worth noting that this increased adoption among women has experienced a slowdown for the second consecutive year, and a notable gender disparity persists.

The uneven representation in digital datasets raises two concerns: the exclusion of unconnected individuals from digital social protection programmes and biases favouring digitally connected groups. To address these concerns, programme designers should integrate digital and traditional models, employing alternative methods like radio, TV, and in-person outreach to reach the digitally disconnected. Non-digital options for decision-making and accessing assistance should also be provided. Costa Rica’s pilot project combines satellite imagery and in-person outreach to ensure inclusive coverage. Using combined approaches can help ensure social protection systems are fair and inclusive, reaching those in need regardless of digital access.

Bias and Discrimination

Algorithmic bias happens when computer programs make unfair decisions based on biases in the data they were trained on, which can affect certain groups of people differently or unfairly. Algorithmic bias in decision-making systems allocating limited resources raises fairness concerns for disadvantaged vulnerable groups, particularly in the context of digital data that may disadvantage those with limited access..

Transparency and explainability are increasingly demanded in data-driven decision-making, including social protection systems. Providing explanations in multiple formats- visual, audio, video, for eligibility decisions, empowers applicants to contest inaccuracies and seek recourse. Offering basic explanations and the option to opt out of private data analysis aligns with informed consent principles. Integrating digital data and advanced algorithms opens possibilities for developing transparent and explainable eligibility criteria that are resistant to manipulation. By analysing data and patterns, algorithms can identify marginalized groups and their information access barriers, enabling targeted outreach strategies and personalized content delivery to bridge the gap and ensure inclusive access for all. Algorithms can also help identify and address the exclusion of certain groups from access to information. Through data and patterns analysis, they can help identify disparities in access and pinpoint areas where interventions are needed, ultimately promoting inclusivity and equal access to information for everyone.

In conclusion, the digitalization of social protection presents immense opportunities for delivering welfare services more efficiently and effectively. However, it also raises critical challenges related to data protection, data poverty, and bias and discrimination. Protecting sensitive data, ensuring privacy through consent, and minimizing surveillance risks are crucial for maintaining trust and upholding individual rights. Addressing the digital divide and providing alternative methods for inclusion will ensure that vulnerable groups are not left behind in the transition to digital systems. Moreover, mitigating algorithmic bias and promoting transparency and explainability in decision-making processes is essential for fair and accountable social protection. By navigating these challenges, social protection programmes can harness the power of digital data while maintaining fairness, inclusivity and ethical standards.[PS7] [MN8] 

Want to learn more? Our interactive and facilitated six-week course, Social Protection for Sustainable Development is designed for anyone interested in learning the basics of social protection and how to apply it through the lens of sustainable development. This course champions a universal, life-cycle approach focusing on partnerships to design, finance, and implement comprehensive social protection systems that reduce the vulnerabilities faced by all.