…

The distinction between technological deficit and multidimensional readiness gap carries significant weight for how development resources might be allocated in fragile states. When international organizations and donor governments approach AI capacity building in contexts like Somalia, the instinctive response often centers on hardware: connectivity projects, data centers, computer labs, and equipment donations. Yet the systematic evidence synthesized in this review suggests that such investments, while necessary, address only one pillar of what determines whether AI adoption translates into productivity gains. A country can possess adequate bandwidth and still fail to benefit from AI if its institutions cannot govern data responsibly, if its universities produce graduates whose skills do not match employer needs, and if no coordinated policy exists to sequence these investments coherently.

The theoretical scaffolding used in the review deserves particular attention from readers unfamiliar with organizational research traditions. The Technology–Organization–Environment framework, originally developed to explain why firms adopt technological innovations at different rates, treats adoption as a function of three contexts: the technology itself, the internal organization, and the external environment. Applying this lens to a nation-state rather than a firm is an analytical extension that proves useful for fragile states, where the organizational context includes not only individual enterprises but the capacity of public institutions to regulate, coordinate, and sustain digital initiatives. Human capital theory, the complementary lens, holds that investments in education and training raise worker productivity, but the Somali case illustrates a critical caveat: such investments only yield returns when the education system produces competencies the labor market actually demands. The review’s finding of persistent curriculum-labor market mismatch demonstrates that human capital accumulation without alignment can coexist with chronic graduate unemployment, a pattern that undermines the simple version of human capital theory and points toward the need for demand-driven skills development.

The paradox at the heart of Somalia’s digital economy merits elaboration. Three decades without a functioning central government produced conditions in which private telecommunications companies filled the void, building network infrastructure and pioneering mobile-money systems that now process transactions at levels remarkable for a low-income economy. This privately led digital dynamism demonstrates that state weakness does not preclude technological adoption; market actors can and do build infrastructure when profit incentives align. However, the same period left public institutions without the regulatory frameworks, data governance capabilities, and coordinated policy apparatus that AI adoption at scale requires. Mobile money thrives on private infrastructure and simple use cases; AI applications in health, agriculture, and public administration demand data sharing standards, privacy protections, quality assurance mechanisms, and interagency coordination that only credible public institutions can provide. The Somali experience thus suggests a nuanced proposition for fragile-state digital development: private-sector-led infrastructure growth is a genuine asset, but it cannot substitute for institutional development, and the two must eventually converge for advanced digital technologies to deliver broad-based benefits.

The evidence base itself constitutes a finding worth examining. The review’s concentration of peer-reviewed research on telecommunications, education, and graduate employment, contrasted with the scarcity of studies on AI readiness in health, agriculture, or public administration, reflects broader patterns in research about fragile states. Researchers gravitate toward sectors where data collection is feasible and where established partnerships exist, which in Somalia has historically meant telecommunications firms and universities in Mogadishu. Agriculture, the country’s largest employment sector, remains understudied despite its vulnerability to climate shocks and its potential to benefit from AI-enabled applications such as weather forecasting, pest detection, and market price prediction. This unevenness means that any readiness framework derived from the literature necessarily rests on partial evidence, and the authors appropriately frame their four-pillar model as a conceptual foundation for future empirical assessment rather than a validated measurement instrument.

The Global Index on Responsible AI scores cited in the review provide a sobering quantitative anchor. Somalia’s scores of 0.26 for both Government Frameworks and Government Actions, ranking 137th, and 0.41 for Non-State Actors, ranking 115th, situate the country at the bottom of global AI governance readiness. These metrics capture more than the absence of an AI strategy document; they reflect the depth of institutional reconstruction still underway after decades of conflict. The finding that over 60 countries have published national AI strategies, with more than 70 percent belonging to developed nations, underscores how concentrated strategic AI planning remains. For Somalia, the absence of such a strategy means that AI-related decisions currently occur in an institutional vacuum, with telecommunications companies, universities, and aid organizations each pursuing digital initiatives without a shared framework for prioritization, coordination, or standards.

The review’s observation about public-sector workforce motivation offers a counterintuitive note of optimism. Empirical assessments finding that Somali government employees are relatively young and express strong motivation to build digital competence suggest that the human raw material for digital transformation exists within the civil service. The obstacle is not attitudinal but structural: training provision is ad hoc, unevenly distributed across ministries, and rarely embedded in a coordinated capacity-building strategy. This distinction matters for policy design. If resistance or apathy were the barrier, awareness campaigns and incentives might suffice; because the barrier is organizational, the remedy lies in systematic training architecture, clear career pathways for digitally skilled civil servants, and ministry-level planning that treats digital competence as core infrastructure rather than occasional enrichment.

The four-pillar framework’s emphasis on labor-market integration as a distinct pillar, rather than a downstream consequence of skills development, reflects the review’s central empirical finding about graduate unemployment. In many countries, expanding higher education eventually produces employment growth as economies absorb graduates. Somalia’s experience demonstrates that this automatic adjustment cannot be assumed in fragile states, where formal private-sector employment remains thin, hiring practices may be non-meritocratic, and career guidance infrastructure is weak. A graduate with digital skills may remain unemployed not because the skills are absent but because the demand side of the labor market cannot absorb them, or because hiring processes fail to match qualified candidates to available positions. Treating labor-market integration as a pillar in its own right forces policymakers to consider employer engagement, internship systems, credential recognition, and hiring reform as integral components of AI readiness rather than afterthoughts.

The phased implementation roadmap proposed for government, higher-education institutions, the private telecommunications sector, and development partners embodies a sequencing logic that deserves attention. Fragile states often face pressure to pursue simultaneous reform across all fronts, an approach that can overwhelm limited administrative capacity and produce fragmented, unsustainable initiatives. A phased approach acknowledges that certain foundations must precede others: basic digital infrastructure and data governance frameworks enable the advanced applications that skills development ultimately targets; institutional readiness enables the coordination that multi-stakeholder initiatives require. The roadmap’s identification of distinct roles for each actor recognizes Somalia’s particular institutional geography, in which the private telecommunications sector possesses capabilities the state lacks, universities produce the talent pipeline, and development partners supply resources and technical assistance, but none of these actors alone can address all four pillars.

The comparative framing with other Sub-Saharan African contexts strengthens the review’s transferability claim. The barriers identified across the continent, including inadequate infrastructure, scarce data ecosystems, weak regulatory frameworks, and shortages of relevantly skilled personnel, are not unique to Somalia, though their severity and configuration vary. Countries such as Rwanda and Kenya have demonstrated that fragile or formerly fragile states can make rapid governance and digital progress when political commitment aligns with sustained investment, suggesting that Somalia’s current position reflects circumstances rather than destiny. The review’s analytical approach, using established frameworks to structure assessment in a data-scarce environment, offers a template that researchers and policymakers in comparable settings, from South Sudan to Yemen to Afghanistan, could adapt to their own institutional conditions.

The methodological choice to conduct a systematic review under PRISMA 2020 guidelines, rather than a narrative synthesis, carries implications for how the findings should be interpreted. Systematic reviews make their search strategies, inclusion criteria, and screening processes transparent and replicable, which matters particularly in a field where advocacy often outruns evidence. The restriction to 31 peer-reviewed sources reflects both the genuine scarcity of rigorous research on AI readiness in fragile states and the review’s commitment to quality over quantity. Readers should understand that this evidence base, while carefully assembled, remains thin, and that the framework’s value lies in organizing future inquiry as much as in summarizing existing knowledge. The authors’ framing of the framework as intended to guide future empirical assessment and policy discussion, rather than as a definitive prescription, is appropriate to this evidentiary situation.

The review’s positioning of AI as a general-purpose technology connects the Somali case to broader debates in economics about how transformative technologies diffuse unevenly. General-purpose technologies, from electricity to computing, historically required complementary investments in infrastructure, skills, and organizational redesign before their productivity effects materialized, often with lags of decades. AI appears to follow this pattern, with firm-level evidence linking adoption to measurable productivity gains particularly where AI augments rather than displaces skilled labor and where workers possess complementary competencies. The augmentation condition is especially relevant for fragile states, where labor displacement without compensating job creation could deepen instability, whereas augmentation strategies that enhance the productivity of existing workers in agriculture, health, and administration could strengthen both economic output and social cohesion.

Finally, the review’s origin within Somali institutions themselves, with authors from Somali National University’s Faculty of Computer Science, the Somali AI Network, and the Dugsiiye organization, represents a meaningful development in the production of knowledge about fragile states. Research about such contexts has historically been authored primarily by external researchers, sometimes with limited grounding in local institutional realities. A systematic review conducted by Somali scholars, published in an international journal, and grounded in the country’s specific conditions signals the emergence of domestic research capacity in exactly the domain, AI and digital skills, that the review identifies as critical. This self-referential quality does not guarantee the findings’ validity, but it suggests that the human capital pillar of the framework is already partially under construction, and that the pathway from readiness assessment to readiness building may begin with the research enterprise itself.

Subject of Research: …

Article Title: Artificial intelligence readiness in fragile states: a systematic review and theory-informed framework for workforce transformation in Somalia

Article References: Jama, A. M., Mohamud, N. F., Hamud, M. O., Hassan, A. M., & Mohamed, A. S. (2026). Artificial intelligence readiness in fragile states: a systematic review and theory-informed framework for workforce transformation in Somalia. Frontiers in Artificial Intelligence, 9, Article 1928521. https://doi.org/10.3389/frai.2026.1928521

Image Credits: AI Generated

DOI: 10.3389/frai.2026.1928521

Keywords: Artificial, intelligence, readiness, fragile, states, systematic, review, theory-informed, framework, workforce, transformation, Somalia