Table of Contents
1. Introduction / Background
The 4th Industrial Revolution (4IR) is firmly underway. Unlike the so-called first three industrial revolutions, this one is firmly powered by AI, the dominant general-purpose technology that has emerged as the engine pushing developments forward.
In common with the other revolutions though, most developments in the technology are emerging in/from developed countries with economies that can support the experimentation which led to these sorts of breakthroughs. In other words, the evidence so far shows that AI emerges in contexts of strong regulatory states, robust capital markets able to bankroll the enterprises pioneering these developments, and mature digital infrastructures.
Incidentally/Ironically/Paradoxically, most leapfrogging examples in history are actually from developing nations, which take a technology painstakingly developed elsewhere and adapted for quick local uptake. Oft-cited examples of this are Somaliland's ZAAD mobile payment system, which substituted banking services with GSM-based accounts for the pre-smartphone era, Kenya's M-PESA which accomplished much the same but also further built on it by making data publicly available, which in turn was used by entrepreneurs to build agricultural tools and services, and Rwanda, where a lack of road and rail infrastructure led to the government partnering with tech companies and using drones for healthcare deliveries.
Of course, the barriers to using GSM-phones and implementing AI-powered services are quite different, as are the challenges presented. Some of these challenges include data — i.e., no common data policies which would provide some levels of protection to individual users.
But this requires adopting, adapting, or rethinking ethics, risk, power, and safeguards. The same digital IDs and AI credit scoring systems that can increase financial inclusion can also reinforce exclusion, surveillance, and algorithmic discrimination — especially relevant in fragile economies where lack of faith in the public sector is often the primary reason they are fragile in the first place.
Encompassing these differences, a core research question can be raised: can fragile economies design technical, institutional, and regulatory systems that would allow them to leapfrog again — this time into AI-enabled service delivery?
Highlighting the accomplished track record of various states in achieving leapfrogging success, the hypothesis of this proposal is that: fragile economies can achieve accelerated development outcomes, especially in public service delivery if AI adoption is structured around contextually adapted data governance models and light-touch public-private partnerships.
The sub-national model
Given the at-times precarious authority of some governments in fragile states, perhaps it would be more prudent to answer this question using a sub-national level of government. If so: what would a subnational model look like (city-level, regional-level), in contexts where national states are weak, fragmented, or not fully internationally recognized?
Somaliland — and specifically the capital Hargeisa — offers a unique case to study. It is functionally self-governing, relatively stable, and already an experimental laboratory for private-sector led digital innovation. Most research focuses on the national level, often ignoring sub-national innovations.
Why Hargeisa? Collect data from the public and private sectors. See their current preparedness. The value they assign to data. An environment where such systems do not currently exist would offer an excellent case in the feasibility of leapfrogging to the 4IR.
2. Research Objectives
The overarching objective of this PhD is to explore how data and technology can drive national development.
The goal of this PhD, and why I would like to do it at a Somali university, is to promote academia as well as closer interactions between the scientific/intellectual and business communities. Therefore, all of the articles are meant to contribute to the public knowledge bank and lead to direct recommendations for businesses and the public sector.
Primary Research Question
How can a fragile/emerging country (i.e., the government of Somaliland — using Hargeisa as a test case) design policies and partnerships to enable private-sector-led AI and data innovation for public service delivery?
3. Literature Review & Theoretical Framework
The following literature review is meant to provide the theoretical framework to be used to conduct the analytical work and derive recommendations. A few core arguments addressing digital leapfrogging, state capacity, and partnerships for public service delivery are presented below, as well as the risks inherent in these contexts and technologies. These theories, arguments, and counter-arguments will be expounded upon during the research program.
Digital Leapfrogging in Weak Institutional Settings
Fragile states often bypass traditional development pathways by adopting disruptive digital technologies. In Somalia/Somaliland, mobile money (e.g., Zaad) replaced collapsed banking systems, demonstrating how institutional voids can spur private-sector innovation. However, most leapfrogging literature focuses on national policies, neglecting the role of subnational actors in driving localized tech adoption.
Subnational Data Governance as Public Good
Weak states lack centralized data infrastructure, but city governments can catalyze AI innovation by treating data as a public good and enable private-sector innovation through open data. For example, Kenya's Open Data Initiative unlocked private-sector AI applications in agriculture. However, existing studies assume state-led data governance, ignoring municipal-level experiments in data sharing.
In Somaliland, Hargeisa could replicate this by releasing, for example, land registry data for property-tech startups, e.g., AI driven blockchain systems that resolve the question of trust plaguing public registries in all fragile markets including Somaliland. Another example could be verified market price feeds for agri-finance algorithms.
Light-Touch PPPs for AI-Enabled Public Services
Public-private partnerships (PPPs) in fragile states can compensate for state weakness, but require minimalist regulation to avoid stifling innovation. Success cases like M-Pesa emerged from sandbox environments where governments provided legal waivers rather than rigid oversight. In Hargeisa, potential PPP models could include AI-assisted healthcare diagnostics tools such as private telemedicine firms using public clinics for patient referral or physical procedures.
On the downside, PPPs in low-governance contexts risk regulatory capture, where private actors dominate public-service delivery and other risks in the form of sensitive data losses or manipulation. A relevant example of that is India's Aadhaar which showed risks of biometrics in low-governance settings. These risk factors must be considered and mitigation measures designed to minimize fallout upon implementation.
Ethical Risks of AI in Fragile Economies
AI adoption in post-conflict zones amplifies algorithmic bias and surveillance risks. For instance, Somaliland's proposed digital ID system could exclude pastoralists lacking biometric records. Or they may not be able to fully enjoy the benefits of any system deployed if there isn't enough data being generated from their behaviour and usage which would allow for targeted solutions, one of the main benefits to be derived from any AI system.
Some mitigation solutions and strategies have been proposed, such as adapt "frugal AI" frameworks that prioritize transparency and offline accessibility. There are also the surveillance risks associated with these artificially intelligent systems, which need to be addressed for a population that is not at all used to digital life (despite having access to digital tools).
The Framework
In summary, one pillar of the theoretical framework under which this research begins is that expecting/enabling leapfrogging using the new, dominant general-purpose technology of the 4th Industrial Revolution, AI, requires new, contextually adapted models. One way of achieving this is by centering on the role of subnational actors (like city governments) in driving localized AI adoption.
The second pillar of this framework involves implementing AI-enabled public services through Public-Private Partnerships (PPPs) which are necessary to compensate for state weakness, with minimalist regulation but with mitigating strategies such as designing measures to minimize fallout from regulatory capture and adopting methods that prioritize transparency and offline accessibility.
4. Methodology & Expected Contribution
This PhD will adopt a mixed-methods approach, combining:
- Qualitative research: Interviews with business leaders, policymakers, and social entrepreneurs.
- Quantitative research: Data analysis of financial models, public service efficiency, and social impact metrics.
- Comparative analysis: Case studies from different national and industry contexts.
To ensure a diverse and well-rounded perspective, data will be collected from Somaliland and at least one other country, depending on the context. This will allow for cross-regional comparisons of data-driven development practices.
Expected Contribution
- A Digital Leapfrog Blueprint for Subnational Governments — Practical policy toolkit for Hargeisa/Somaliland, adaptable to Somalia and other fragile states in Africa and the Global South. This would also include recommendations for businesses and social enterprises on leveraging data for national development.
- New Theories on Decentralized AI Governance — How "weak" states can still enable tech growth via city-level innovation.
- Real-World Testing — Via PPP pilots in Hargeisa (e.g., AI for property taxes, chatbot-based business licensing).
I would like to publish the results of my research in the form of a series of papers.
The dissertation will be article-based with three planned peer-reviewed publications:
- Article 1: Municipal Data as a Public Good in Fragile Cities: The Hargeisa Case — Can city-level open data policies in fragile states catalyze AI solutions for public services?
- Article 2: Light-Touch PPPs for AI-Enabled Public Services: Design, Risks, and Results
- Article 3: Ethical Guardrails and "Frugal AI" for Inclusion in Post-Conflict Settings
All findings will be synthesized into a book with policy recommendations.
