|
Listen to story
Getting your Trinity Audio player ready...
|
This article exposes the vulnerability of unregulated deployment of artificial intelligence to political capture in semi authoritarian regimes in sub–Saharan African governments. Secondly, the piece holds the government of Zimbabwe to account for the digital rights violation to data privacy and explainability of the covid 19 relief grant algorithm following allegations of corruption and unfair distribution of social goods.
The African Charter for human and Peoples Rights (ACHRP) is working flat out to maximize benefits that come with artificial intelligence and to minimize harm against the background of human rights.
Following the recommendations by the ACHRP resolution of 2021 which implored states to undertake studies on AI and emerging technologies, majority of the African countries have not come up with AI strategies and policy which has left room for the deployment of AI in unregulated markets resulting in violation of digital rights to data privacy.
It is the digital rights call, that governments and private technology companies must be held accountable for the deployment of AI and the ethical implications that come with emerging technologies.
In this article, I unpack the ethical implications of deploying artificial intelligence in African countries without existing policy on emerging technologies against the background of digital rights to privacy. Secondly, I will look into how digital poverty is a function of social injustice during against the background of deployment of emerging technologies in sub–Saharan Africa. Lastly, will be the explainability of the algorithm which raised concerns on the issue of accountability following the irregularities and discrepancies during the distribution of the funds.
Just like any other country, the government of Zimbabwe declared the novel covid19 pandemic a national disaster in March 2020. Immediately, the ministry of finance released 6 million USD as disaster relief fund to cushion poverty-stricken communities during the covid 19 induced catastrophe with the country under lockdown.
Supported by good intentions, assigning artificial intelligence in the distribution of cash-based transfers had benefits resulting in faster, cheaper and simpler distribution of the funds and also allowed for monitoring and tracking of the funds. However, what followed next in the distribution of the Covid19 relief grant exploded into social injustice grand scale corruption and scandal.
Zimbabwe’s ministry of information announced that government had acquired a “sophisticated algorithm”, to distribute the covid19 relief grant using the services of government linked mobile operator, Netone through its mobile money service One Wallet. This decision by the Zimbabwean government presented elements of unequal distribution of social goods during time of disaster let alone the rights to data privacy.
The deployed algorithm required bank balance, mobile money wallet balance and location as variables to determine legibility for the relief. Central to this discussion is how data subjects ( The intended beneficiaries ) did not give consent data access to third parties.
I can hypothetically assert that, data collected led to politically influenced decisions to appropriate the funds ruling party sympathisers considering that the distribution partner, Netone only had a market share of 12% among other telecommunications service providers in Zimbabwe 2020.
The problematic algorithm
The so called “sophisticated algorithm” operated on three variables : Bank account balance, mobile money wallet balance and location based on the data provided by the network service provider. Technically, the algorithm was deployed on the assumption that most vulnerable communities would meet the criteria of the variables.
To start off, investigative media reports revealed harrowing evidence of possible corruption and technology failure when more than 15 people living in the same household were picked by the algorithm for eligibility. This evidence is corroborated Belliveau (2016) in the extract below:
“that over-dependence on technology might possibly result in the erosion of the practical comprehension and compassion that should direct and inspire the humanitarian response as there will be a wide divergence between the aid workers and those in need. For example, Scott-Smith (2016) posed that bunkerisation prevents aid workers from interacting with aid recipients thus they may lose good awareness of the societies with which they are dealing”
Another problematic area with the algorithm which paved the way for collusion and corruption is that government approved that vulnerable Zimbabweans had access to bank accounts and mobile phones. With 67 % of the Zimbabwe’s population living in rural areas without access to banking and mobile wallet transactions. Rural communities are still accustomed to a cash and batter system. Moreso, with an unemployment rate of over 80% among the youth, the 33% of Zimbabweans living in urban areas do use the bank as they are overly reliant on the informal sector for economic activities.
Explainability of the algorithm
To date, the artificial intelligence deployed by the government to distribute 6 million USD relief grants is still unexplainable following the algorithmic failure highlighted by the auditor general report 2021, that over 1 million USD was lost to duplicate and unclaimed transfers. This problem was further expanded by Kudakwashe Mugoto : (Algorithms in the distribution of Covid 19 relief funds in Zimbabwe ), in the extract below:
“For example, Findings from the Auditor Generals (2021) report highlights those 58 recipients in Mutare who received COVID-19 allowances amounting to ZWL$45 240 had same identity numbers, different dates of birth and different gender while over 375 recipients had uncontactable addresses, suspicious names and identity numbers and received COVID-19 allowances amounting to ZWL$292 500. These anomalies then point to a situation whether the algorithm was ever used, or this was a sophisticated way to cover up misuse of funds”.
Moreso, the government explained a three-stage process of the cash transfers which was challenged by Mugoto to have been a six-step stage that paved way for influence by human administrators that may have been involved in the compilation of lists submitted for cash distribution.
The stages are illustrated in in the diagrams below.
Pictures
Source :Algorithms in the distribution of Covid 19 relief funds in Zimbabwe )
Evidently, explainability of the algorithm is questionable if it is considered that Mugoto’s hypothesis highlights creation of lists from the stage of data entry before the algorithm is used to determine recipients and the creation of a final list before a list is sent to Netone the distribution partner.
The digital rights concerns
Though Zimbabwe had not enacted the Cyber and Security Act of 2021 which prohibits access to data by third parties without the consent of data subjects. In the case of Zimbabwe where a mobile operator had access to bank account as illustrated in the extract below:
Media Institute of Southern Africa and Collaboration on International ICT Policy for East and Southern Africa’s CIPESA (2021:9) realisation that “the use of the algorithm highlights the extent of privacy threats in Zimbabwe as the data subjects were neither advised nor required to give their consent to the sharing of personal data for the relief programme.”
The deployment of the algorithm created a privacy concern which could have resulted in the government facing legal challenges as personal details of people were going to be collected from the mobile providers was going to be without consent of the general populace and in violation of section 57 of Constitution of Zimbabwe that guarantees the right to privacy.
The government assumed that everyone had access to a mobile phone let alone a Netone sim card. Though, POTRAZ, was reporting an increase in mobile phone penetration but if it is considered that people living in urban areas can own more than two active Sim cards, it then questions the accuracy of the reports as reliable information.
The operating nature of the algorithm presented digital rights concerns on data privacy. Firstly, a third party, had access to process and access banking and location data of people without consent of the data subject.
The second problematic area is that the algorithm relied on data provided by network service providers presented inequality considering that vulnerable Zimbabweans who did not own a mobile phone at the time were technically excluded from the covid 19 relief grant as result of disqualification by the algorithm.
Rounding up
This article is one of the many publications challenging African governments and private tech companies to account for the AI deployment in unregulated markets. It is imperative to consider that ethical implications of emerging technologies have a negative bearing on human rights. Central to the violations is the right to data privacy, where corporates may illegally access personally identifying data and information to fuel advertising campaigns. In another dimension, unregulated deployment of AI paves the way for corruption and embezzlement public and social goods leaving the most vulnerable communities at the mercy of politicians in decision making.
If you support independent journalism in Zimbabwe, you are welcome to donate towards ZNCJ.ORG fund.