Martha Chukwu
Title: ALGORITHMIC GOVERNANCE IN NIGERIA’S HEALTH CARE INDUSTRY – UBENWA AI SOLUTION
Supervisor: Prof. Dr. Anu Masso, Dr. Stefano Calzati
Opponent: Dr. Egert Juuse
Defense: 17 August 2020
Abstract: Big data has revolutionized human activities and their impact across many sectors, including healthcare. Privacy, safety, security, and human rights concerns remain heightened despite the many values it brings. Different jurisdictions are addressing these concerns differently, notably by developing guidelines and regulations. Social datafication, governance through data, data colonialism, and interoperability governance are emerging areas of research interest. Though behind, Africa is increasingly datafied with regulatory gaps whose potential can result in re- enforced inequalities and biases. Localized data are often unavailable/ almost non-existent in Africa. Hence, digital solutions deployed in the continent are developed with data from a different social context, causing social inequality and bias. Ubenwa, a use case AI solution that uses the cry of a baby to detect birth Asphyxia, was analyzed to understand how discrimination and inequality can occur from its design and interoperability, how to avoid unintended social harm and how social context plays a role in algorithmic governance. Data governance frameworks were evaluated. In-depth qualitative interviews were conducted with a cross-section of data subjects, who will be impacted by such an AI solution; and health data experts who will use such a solution. This research also adopted an online quantitative survey method to ascertain the state of data and AI governance in Nigeria, and understand the level of awareness and compliance in addressing algorithmic governance, data colonialism, and interoperability governance. The literature shows that while there are international frameworks for ethical AI, none exists in Nigeria. The interview and survey analysis results show that algorithmic governance is timely. The study found that its successful implementation is dependent on contextualizing initial algorithmic dataset, and regulating and enforcing its use with frameworks/ guidelines. The study made recommendations for improving and supporting algorithmic decision making in the country.
Keywords: Social datafication, Data colonialism, Algorithmic governance, Data-driven / Data- informed governance, Data ethics, Interoperability, Artificial Intelligence, Big Data.
