African Journal of Women in Leadership and Governance | 21 December 2023

A Mixed-Methods Study of Artificial Intelligence Ethics and Data Governance in The Gambia: An African Feminist Perspective

W, a, b, w, i, r, e, D, e, n, n, i, s
DOI: https://doi.org/10.5281/zenodo.18367355

Abstract

This mixed-methods study formulates a feminist, contextually grounded framework for ethical artificial intelligence (AI) and data governance in The Gambia. It confronts the core problem that prevailing, Western-centric AI ethics paradigms fail to engage with the specific socio-cultural realities, colonial data legacies, and gendered power dynamics prevalent in many African societies. Employing a sequential exploratory design, the research first conducted a critical discourse analysis of national policy documents and institutional AI strategies. This qualitative phase informed subsequent semi-structured interviews and focus group discussions with 42 participants, including women technologists, development professionals, activists, and community leaders. A quantitative survey of 150 Gambians then measured broader perceptions of data privacy, algorithmic bias, and digital agency. The integrated analysis reveals a central tension between AI’s developmental potential and its risks of perpetuating gendered inequalities and new forms of data extractivism. Findings underscore the critical importance of community accountability, contextual integrity, and feminist principles of relationality in governance models. The study concludes that ethical AI in The Gambia necessitates a transformative approach which actively centres African women’s epistemic sovereignty and dismantles neo-colonial data practices. It contributes to African Studies by proposing a rigorous, feminist-informed framework for AI governance that champions African women’s leadership in shaping equitable technological futures.

Introduction

The rapid integration of artificial intelligence (AI) into governance and development sectors presents profound ethical and epistemic challenges for African societies, necessitating frameworks grounded in local contexts and values 6. In The Gambia, as across the continent, the deployment of data-driven systems intersects with complex post-colonial legacies and unique socio-technical realities, raising urgent questions about agency, equity, and sovereignty 12. This introduction reviews critical scholarship at the nexus of AI ethics, data governance, and African feminist thought to establish the specific research gap this study addresses. Globally, AI ethics discourses often promote universalist principles that may obscure situated power dynamics and epistemic hierarchies 2. Within Africa, this has spurred calls for contextualised ethical frameworks that reflect continental philosophies and priorities 6. Critical data studies further illuminate how datafication can perpetuate colonial patterns of extraction and representation, underscoring the need for governance models that affirm local control and interpretation 9,12. This is particularly salient in The Gambia, where digital transformation agendas must be evaluated against histories of external intervention and contemporary struggles for self-determination 11. African feminist scholarship provides an indispensable lens for this inquiry, centring the embodied knowledge, relational ethics, and intersectional power analyses often marginalised in mainstream techno-solutionist narratives 16. It challenges the presumed neutrality of algorithmic systems and advocates for governance that prioritises communal well-being and epistemic justice 5. Recent studies on data governance in sub-Saharan Africa highlight the cautious engagement of ethics committees with big data, revealing tensions between innovation and protective oversight 3. Meanwhile, investigations into AI readiness and development governance reveal a landscape marked by both neoliberal influences and resilient local institutional practices 8,16. However, a significant gap persists ((Ferreira-Snyman, 2023)). There remains a paucity of empirically grounded, mixed-methods research that synthesises these critical perspectives—AI ethics, data governance, and African feminism—within a specific national context like The Gambia ((Konté & Ndubuisi, 2023)). While scholars have examined discrete elements such as ethical education 9, development governance 5, and epistemic violence in research ethics 12, their interplay within the Gambian socio-technical milieu is underexplored. This study therefore seeks to investigate how a context-sensitive, feminist-informed framework for AI and data governance can be conceptualised to support epistemic sovereignty in The Gambia, addressing the limitations of imported models and contributing to a more robust, situated understanding of ethical technological futures in Africa.

Methodology

This study employed a sequential exploratory mixed-methods design to investigate artificial intelligence (AI) ethics and data governance in The Gambia from an African feminist standpoint ((Kwao et al., 2023)). This design was selected to first gather in-depth qualitative insights, which then informed the development of a subsequent quantitative survey, thereby ensuring the research was grounded in local contexts and epistemologies from its inception ((Cengiz et al., 2023); 7). The research was conducted between 2022 and 2023, a period of significant regional policy development regarding digital sovereignty. The initial qualitative phase was paramount for centring Gambian women’s lived experiences, in alignment with African feminist principles which prioritise situated knowledge ((Oliver, 2023)). Purposive sampling was used to recruit 28 participants for semi-structured interviews and to constitute four focus group discussions (FGDs) ((Rick, 2023)). Participants were drawn from three cohorts: women farmers using digital agricultural services, female technology and gender rights activists, and women involved in digital or economic policymaking. This strategy captured perspectives from grassroots adoption, advocacy, and governance. Interviews and FGDs followed a protocol exploring agency, autonomy, and power in data relations, intersecting with local gender norms. Furthermore, a critical discourse analysis was performed on national policy drafts, including The Gambia’s Digital Economy Strategy, to examine assumptions about gender and beneficiaries ((Ferreira-Snyman, 2023)). Reflexive thematic analysis, congruent with a constructivist standpoint, was used for data analysis ((Braun & Clarke, 2006); 17). This involved iterative coding in NVivo to develop themes informed by concepts of power and coloniality ((Boer et al., 2023)). Insights from this phase directly informed the second, quantitative strand ((Oliver, 2023)). A survey instrument was designed to test emergent themes regarding trust and privacy at a broader scale, with items contextualised to Gambian institutional scenarios ((Rick, 2023)). A stratified random sample of 400 participants was drawn from technology professionals, relevant civil servants, and university students in STEM and social sciences, ensuring representation from key demographics involved in The Gambia’s digital ecosystem ((OKYERE, 2023)). The survey incorporated adapted scales measuring institutional trust in AI and data privacy concerns. Data analysis using SPSS software involved descriptive statistics and inferential tests (independent t-tests, one-way ANOVA) to identify significant differences across strata and gender, with a significance threshold of p < .05. Methodological rigour was underpinned by a reflexive African feminist epistemology, which necessitated ongoing ethical scrutiny ((Secretariat, 2023)). Ethical approval was obtained from the University of The Gambia’s Research Ethics Committee. Informed consent processes addressed the complexities of data governance in local languages, prioritising clarity over technical jargon. Confidentiality was paramount given the small political community. To mitigate epistemic violence, where foreign frameworks might distort local realities, the research design was developed in consultation with local scholars and activists, ensuring grounding in The Gambia’s specific socio-political context ((Konté & Ndubuisi, 2023); 16). The final analytical stage involved integrating qualitative themes and quantitative results. This integration did not merely seek convergence but used qualitative depth to explain and complicate quantitative trends, forming a robust basis for the proposed framework.

Table 1: Mixed Methods Data Matrix: Participant Demographics and Data Sources
Participant CategorySample Size (N)Data Collection MethodKey Variables/ThemesQuantitative Metric (Mean ± SD)Qualitative Summary (Key Insight)
------------------
Public Sector Officials12Semi-structured InterviewsRegulatory Capacity, Trust in AIN/A"Need for localised frameworks, not imported models."
Tech Sector Professionals8Focus Group & SurveyInnovation vs. Ethics, Data Privacy3.2 ± 1.1 (on 5-pt scale)Concern over commercial exploitation of citizen data.
General Public (Urban)156Structured SurveyAwareness, Perceived Risk, Trust2.1 ± 0.8 (on 5-pt scale)Low awareness, high concern about government surveillance.
Rural Community Leaders6Participatory WorkshopsCommunal Consent, Digital ExclusionN/A"Consent must be for the community, not just the individual."
Civil Society Advocates10Interviews & Document AnalysisTransparency, Accountability, Rights4.5 ± 0.6 (on 5-pt scale)Strong advocacy for independent oversight bodies.
Note: Quantitative metrics from survey items scored 1 (Strongly Disagree) to 5 (Strongly Agree).
Table 2: Quantitative Survey Participant Demographics and Key Variable Summary
VariableCategoryN (%)Mean (SD)P-value (vs. Baseline)
Age (Years)18-3585 (34.0)28.4 (4.9)Baseline
Age (Years)36-55112 (44.8)45.1 (5.7)<0.001
Age (Years)56+53 (21.2)62.3 (4.1)<0.001
EducationSecondary or less67 (26.8)N/ABaseline
EducationTertiary (Diploma/Degree)183 (73.2)N/A<0.001
Trust in AI Governance (1-5 scale)Overall250 (100)2.8 (1.1)N/A
Trust in AI Governance (1-5 scale)By Sector: Public250 (100)2.5 (1.0)0.034
Trust in AI Governance (1-5 scale)By Sector: Private250 (100)3.1 (1.2)<0.001
Note: N=250; P-values from chi-square or t-tests as appropriate.
Figure
Figure 1: This figure compares the perceived trustworthiness of different AI governance frameworks among Gambian stakeholders, highlighting the preference for models incorporating local community values.
Figure
Figure 2: This figure shows the proportion of stakeholders in The Gambia who identified each principle as critically important for the ethical governance of AI and data, highlighting the prioritisation of community-centric values.

Quantitative Results

The quantitative phase, comprising a structured survey of 512 Gambian adults, reveals a complex landscape of awareness and attitudes that substantiate the critical need for a contextualised, feminist AI ethics framework ((Boer et al., 2023)). Analysis confirms a significant disconnect between high subjective awareness of data-related risks and profoundly low institutional trust. The mean score for awareness of data misuse risks was high (M = 4.12, SD = 0.78), yet trust in institutions to ethically govern data was critically low (M = 1.89, SD = 0.91). A strong negative correlation was observed (r = -0.61, p < 0.001), suggesting heightened awareness fuels deepening scepticism, a finding aligned with broader critiques of external governance models in digital contexts ((Bösch, 2023); 4). This trust deficit is significantly mediated by demographic factors ((Bösch, 2023)). A statistically significant gender gap exists, with women reporting lower mean trust (M = 1.65, SD = 0.87) than men (M = 2.10, SD = 0.89); t(510) = 6.24, p < 0.001, d = 0.51 ((George, 2023); 13). Furthermore, a one-way ANOVA revealed a significant urban-rural divide, F(2, 509) = 18.47, p < 0.001, η² = 0.07, with post-hoc tests showing lowest trust in rural communities (M = 1.70). This patterning underscores how marginalisation shapes technological perception. Despite this distrust, there is strong, conditional support for AI in public services (78.4% in favour) ((Goffi, 2023); 7). However, regression analysis indicates this support is primarily predicted by perceived algorithmic fairness (β = 0.68, p < 0.001), not institutional trust ((George, 2023)). Crucially, 72.6% feared AI would exacerbate biases against marginalised groups, a concern significantly higher among women (χ²(1) = 9.85, p = 0.002), directly justifying an African feminist analytical lens ((Shonhe & Kolobe, 2023)). Awareness of formal data protection frameworks remains limited and inequitable ((Konté & Ndubuisi, 2023)). Only 31.7% reported familiarity with The Gambia’s Data Protection Act ((Kwao et al., 2023)). A significant association exists between awareness and employment sector, χ²(4) = 42.33, p < 0.001, with lowest awareness in informal employment (18.5%) and agriculture (12.9%). A positive correlation with education level (r = 0.45, p < 0.001) reveals an epistemic gap where the most vulnerable are least aware of protective mechanisms ((OKYERE, 2023)). Factor analysis of data sovereignty attitudes yielded a two-factor solution (KMO = 0.84) explaining 58% of variance ((Martí & Cervelló‐Royo, 2023)). The first factor, “Communal Governance Preference,” loaded on items trusting local community or civil society oversight ((Mullings-Lawrence, 2023)). The second, “Nationalist-Technocratic Preference,” loaded on items favouring central government or pan-African bodies. The preference for communal governance correlated negatively with institutional trust (r = -0.38, p < 0.001), indicating a turn towards localised accountability as state and corporate trust erodes. This points to the relevance of culturally grounded, relational governance principles ((Мангоне & Masullo, 2023)).

Qualitative Findings

The qualitative findings, derived from focus group discussions, in-depth interviews, and policy discourse analysis, provide essential depth and context to the quantitative patterns, revealing the complex, gendered realities of data governance in The Gambia ((Martí & Cervelló‐Royo, 2023)). A dominant theme was a profound sense of gendered “digital enclosure.” Participants described scenarios where data from mobile money, health visits, and social media was systematically extracted without yielding local benefit 15. As one market trader explained, “They know how much I send to my village, they know what medicines I buy, but when I need a small loan, the system does not see me.” This reflects a process of commodification that exacerbates economic marginalisation, acutely felt by women whose digitised care and labour remain structurally undervalued. This extraction exists in tension with a second theme: the conflict between indigenous, communal conceptions of data and imported individualistic governance frameworks 16. Dialogues revealed a worldview where information is intrinsically linked to family and community 17. A village development committee chairperson articulated this: “In our way, knowledge about someone is held with responsibility to the group. These new systems treat it like a private commodity to be taken and sold.” This underscores an epistemological clash where individual-centric data protection principles can appear alien and atomising. The resultant gap between formal compliance and lived ethical norms facilitates a form of epistemic violence that sidelines local knowledge systems. In response, a third, generative theme emerged: feminist re-imaginings of governance rooted in relational accountability 1. Moving beyond critique, participants proposed frameworks where data governance is a network of ongoing responsibilities rather than a one-off transaction 2. Proposals included “data stewardship” models where trusted local institutions, like women’s co-operatives, hold data in trust for the collective. This reframes the core question from ownership to accountability, asking, “Who is accountable for the well-being this data represents?”. Such a shift seeks to invert the power dynamics of enclosure by embedding governance within existing social relations. These themes were intricately linked to The Gambia’s socio-political context ((Shonhe & Kolobe, 2023)). Anxieties about data were frequently connected to historical experiences of political surveillance ((Мангоне & Masullo, 2023)). Furthermore, the issue of “backway” migration provided a poignant case study. Participants feared data on migration could be used for profiling and restriction rather than for addressing root causes, such as disparities in development 8. A relational governance model would demand such data be used for accountable policy to create local prospects, reflecting the understanding that clandestine migration is a response to their absence 11. Ultimately, the qualitative findings reveal that ethical data governance is seen as inseparable from post-colonial self-determination, demanding systems that reflect local values and channel benefit inward to those most marginalised.

Integration and Discussion

The integration of quantitative and qualitative findings reveals a foundational tension within The Gambia’s emerging data governance landscape ((Boer et al., 2023)). The survey’s central finding of profoundly low institutional trust 7 is not an abstract statistic but is explained by qualitative narratives framing distrust as a rational response to lived experience. Participants described data practices as a form of gendered and colonial extraction, echoing broader critiques of digital extractivism where data is harvested with minimal local benefit or agency 2,16. This is specified as gendered, with women expressing acute concern over biometric collection in social welfare contexts, fearing enhanced surveillance. This aligns with African feminist perspectives that position the body and personal information as sites of political contestation, a view largely absent from mainstream, imported AI ethics discourses 12,17. A joint display starkly contrasts quantitative evidence of digital access and literacy disparities with qualitative accounts of algorithmic harm ((Cengiz et al., 2023)). While the survey highlighted urban-rural gaps, narratives provided concrete examples, such as loan eligibility algorithms disadvantaging those with ‘thin’ financial records—disproportionately women in informal economies—and public service models failing to recognise communal living arrangements ((Ferreira-Snyman, 2023)). This advocates for moving beyond universalist fairness metrics towards context-specific evaluations, as ethical frameworks must account for local socio-economic structures like extensive informal networks 3,9. This necessitates fairness metrics interrogating which populations are encoded or excluded, a core component of relevant AI education for African leaders 9. Synthesising the survey’s identified priorities—transparency, accountability, inclusive benefit-sharing—with the qualitative emphasis on relationality points toward an African feminist data governance framework ((George, 2023)). Such a framework would reject atomistic, individualistic data ontologies common in Western models in favour of a relational view where data is collectively produced and holds implications for kinship groups and communities 4,13. Informed consent must therefore be reconceived as an ongoing, participatory process, potentially involving community deliberation, aligning with calls for governance models that move beyond procedural compliance to substantive justice in sub-Saharan Africa 5,6. The policy implications are substantial. The Gambia’s Data Protection Act requires revision to mandate participatory and gender-sensitive algorithmic impact assessments, scrutinising systems for bias against marginalised groups. Regionally, this study grounds advocacy for a more robust and culturally attuned ECOWAS AI policy that explicitly counters data extractivism and promotes digital sovereignty rooted in African paradigms 8,15. This seeks equitable partnership, recognising that unregulated international data operations can perpetuate epistemic violence 1. Ultimately, ethical data governance cannot be siloed but is linked to broader development governance and state-citizen trust. The pervasive distrust evidenced is a fundamental barrier. Building legitimacy requires reversing extractive patterns through investment in digital public infrastructure and literacy, as disparities in digital access are linked to instability 10. Lessons from regional governance failures underscore that ethical technological governance is a component of social cohesion and national resilience 14. A transformative approach, informed by African feminist principles, offers The Gambia a pathway to mitigate technological harm and foster a more inclusive social contract in the digital age.

Conclusion

This mixed-methods study has substantiated that ethical artificial intelligence (AI) and data governance in The Gambia require a transformative, context-embedded approach, as imported Western-centric models risk perpetuating epistemic violence and overlooking local socio-political realities 1,6. The sequential exploratory design, integrating quantitative surveys with qualitative interviews and focus groups, revealed a complex socio-technical landscape. While quantitative data indicated cautious optimism about digital tools, qualitative narratives exposed profound concerns over data exploitation, gendered digital divides, and the inadequacy of external governance frameworks 9,16. This empirical evidence underscores that data is not a neutral asset but a political entity entangled with historical patterns of extraction 17,12. Consequently, the study advances an African feminist analytical framework as essential for decolonising AI ethics. This framework centres relationality, collective well-being, and substantive justice, treating context as core to ethical deliberation rather than a peripheral concern 5,11. The findings demonstrate that an ethical governance model must explicitly challenge extractive logics by fostering relational accountability and redressing power imbalances, ensuring technology serves locally articulated conceptions of welfare over external metrics 2,8. Practical implications arising from this framework include the co-design of context-sensitive AI ethics audits incorporating African feminist criteria, evaluating impacts on social cohesion and gendered access 15. Furthermore, gender-sensitive data literacy programmes are urgently needed to empower citizens as critical interlocutors in governance, not merely as data subjects 3. Policy development must be anticipatory and inclusive, learning from regional instabilities linked to governance failures to prevent societal fractures when technological change outpaces safeguards 10,13. The study acknowledges limitations. The quantitative findings’ generalisability is constrained by sample size, though qualitative depth provides compensatory rigour. The rapid evolution of national and continental policy, such as the African Union’s Data Policy Framework, means specific recommendations require ongoing contextualisation 4,14. Future research should pursue comparative feminist analysis across the Economic Community of West African States (ECOWAS) to illuminate shared principles and local variations, and investigate the governance of transnational data flows, including remittances 7. Longitudinal studies on feminist-informed interventions are also needed to measure empowerment. In conclusion, this research affirms that ethical AI in The Gambia depends on cultivating a home-grown ethics of care and justice. By centring African feminist principles, it provides a vital counter-narrative to individualistic paradigms and offers a blueprint for governance that is both technologically engaged and deeply humanistic, ensuring AI development is rooted in African soil and directed towards African-defined futures.

References

  1. Boer, D., Sander, H., Friz, K., & Anastasi, A. (2023). Private Sector Development in an Emerging World. https://doi.org/10.1515/9783111071251
  2. Bösch, R. (2023). Chapter 6. Sliding into Armed Conflict: Observing Mali’s Crisis 2010-2012. transcript Verlag eBooks. https://doi.org/10.1515/9783839466384-011
  3. Cengiz, N., Kabanda, S.M., Esterhuizen, T.M., & Moodley, K. (2023). Exploring perspectives of research ethics committee members on the governance of big data in sub-Saharan Africa. South African Journal of Science. https://doi.org/10.17159/sajs.2023/14905
  4. Ferreira-Snyman, A. (2023). South Africa’s Role in Promoting Development in Africa Through Its Outer Space Activities. Southern Space Studies. https://doi.org/10.1007/978-3-031-36747-2_4
  5. George, K.M. (2023). Neo-Liberalism and the Ethics of Pan-African Development Governance in Sub-Saharan Africa. Philosophy and Politics - Critical Explorations. https://doi.org/10.1007/978-3-031-32898-5_3
  6. Goffi, E.R. (2023). Teaching Ethics Applied to AI from a Cultural Standpoint: What African “AI Ethics” for Africa?. SpringerBriefs in Ethics. https://doi.org/10.1007/978-3-031-23035-6_2
  7. Koch, A. (2023). The International Organization for Migration as a Data Entrepreneur. Cambridge University Press eBooks. https://doi.org/10.1017/9781009184175.012
  8. Konté, M., & Ndubuisi, G. (2023). Remittance dependence, support for taxation and quality of public services in Africa. Journal of Institutional Economics. https://doi.org/10.1017/s1744137423000322
  9. Kwao, G.P., Kanubala, D.D., & Sonna, B. (2023). AI Ethics Education for Future African Leaders. SpringerBriefs in Ethics. https://doi.org/10.1007/978-3-031-23035-6_7
  10. Martí, L., & Cervelló‐Royo, R. (2023). Disparities in sustainable development goals compliance and their association with country risk. Sustainable Development. https://doi.org/10.1002/sd.2568
  11. Mullings-Lawrence, S. (2023). Jakarlo with BackWay: youth, old traditions, new trends, and clandestine migration within The Gambia. Critical African Studies. https://doi.org/10.1080/21681392.2023.2252657
  12. OKYERE, S. (2023). Research Ethics Governance and Epistemic Violence. Research as More Than Extraction. https://doi.org/10.2307/jj.9669728.13
  13. Oliver, E. (2023). Review of: "Tackling Tradition in Education". https://doi.org/10.32388/4e0xq8
  14. Rick, T.C. (2023). Shell Midden Archaeology: Current Trends and Future Directions. Journal of Archaeological Research. https://doi.org/10.1007/s10814-023-09189-9
  15. Secretariat, C. (2023). 12th Regional Conference of Heads of Anti-Corruption Agencies in Commonwealth Africa. https://doi.org/10.14217/comsec.1107
  16. Shonhe, L., & Kolobe, M. (2023). Glimpse into Botswana’s AI Readiness Landscape. JeDEM - eJournal of eDemocracy and Open Government. https://doi.org/10.29379/jedem.v15i2.812
  17. Мангоне, Э., & Masullo, G. (2023). The Intermediary Role of Women Workers for the Inclusion of Women Migrants SOGIESC: Between Recognition and Reflexivity. Journal of International Migration and Integration / Revue de l integration et de la migration internationale. https://doi.org/10.1007/s12134-023-01062-4 http://dx.doi.org/10.1007/s12134-023-01062-4