Ко добија право да гради вештачку интелигенцију? Решавање недостатака у инфраструктури, подацима и управљању.
Интервју са Филипом Тигом, специјалним изаслаником за технологију у Кенији
The interview was conducted by Ula Rutkowska (Senior Researcher, Apolitical) and edited by Christina Obolenskaya (MSc in International History, LSE and Communications Intern, Apolitical). Phillip Thigo was named on Apolitical’s Government AI 100 2025.
Artificial intelligence is often framed as the domain of a handful of well-resourced nations and corporations, but its potential to transform governance, public services, and economic development is global. For countries in the Global South, however, AI adoption faces significant hurdles—gaps in infrastructure, data access, digital skills, and governance models make equitable AI development a challenge.
Phillip Thigo, Kenya’s Special Envoy on Technology, explores what it takes to build the foundational systems needed for AI to be rolled out at the global level. From investing in energy and connectivity to rethinking public-private partnerships and financing models, he highlights the steps necessary to ensure AI benefits are shared beyond a few powerful economies. Thigo argues that inclusivity, trust, and cooperation must be central to AI’s future—without them, the technology risks deepening existing inequalities rather than addressing them.
Q: What are the most pressing challenges in creating equitable AI policies, particularly for nations in the Global South?
Foundations of AI — AI is often seen as part of a higher echelon of innovation and technology, involving data, computing, and algorithms. However, before countries can deal with these advanced functions, they need to invest in basic foundations to ensure equitable access to these advanced capabilities. For instance, you need Data to train AI, whether it’s in health, agriculture, or education. The global majority has significant data gaps that are fundamental to addressing their development challenges such as the Sustainable Development Goals.
Infrastructure — More specifically, internet connectivity. The continent is still significantly under-connected. For example, in Eastern Africa, only 24% of people are connected to the internet, while Kenya's connectivity rate is at 42%. This highlights the fundamental issues around infrastructure, that also includes energy. AI consumes a lot of energy, yet the African continent faces a significant energy deficit, with over 600 million people lacking access to electricity, and moreso last mile connectivity, hindering economic growth and social development.
Scalable skills — Basic skills around STEM, from primary education, high school, tertiary and university. The continent further lags behind in industry specific skills that are necessary for driving the new AI Economy.
Agile governance — We need to figure out how to govern these technologies in a way that enables development, while also addressing its risks and challenges. The continent requires strengthening its governance mechanisms, which must include a whole of government: parliamentarians and judiciary who have a primary role in legislation and arbitration.
Innovative financing — Many countries in the south are fiscally constrained, and will require to re-think investments in the digital economy without falling into further debt. Finally, the continent requires partnerships that are equitable as the current model of private-public partnerships are not working. It is just not possible to partner with a company worth $3 trillion, potentially backed by the source country when your GDP is less than $100 billion.
While we talk about compute, data, talent, and algorithms, we must also not forget the basics as inclusivity must remain a critical principles in this new economy. Not many countries can afford compute capacity and it does not make economic sense for everyone to invest in this infrastructure. Therefore we need to think about potential sharing of these large scale investments, while at the same time, countries should invest in the foundations so that their citizens can benefit from this new economy. Kenya, for example, is well-positioned due to our clean energy, broadband connectivity and robust technology ecosystem, but not sufficient workload for compute. How do we therefore ensure that neighboring countries can leverage on this capability within their sovereignty? How therefore, do we develop a cohesive regulatory regime that enables cross border data flows that guarantees safety, security and protection of sovereign data.
Q: With a fast-moving technology like AI, what is the path to building these foundations to ensure equity in AI development?
Any AI-rich nation requires the same foundations. We have recently seen announcements around investments in AI, whether the $500 Billion announcement by the White House and OpenAI or the AI Action Summit in France, Canada and the European Union. You realise that the key to AI adoption is investing in Infrastructure. What this requires therefore is the need to reinvigorate our development architecture that understands the need for investing in technology akin to rail, roads and waterways.
While countries like Kenya need to double down on investments, it also requires a serious re-evaluation and reform of the current financial architecture that enables access to fair credit facilities and repayment.
As previously noted in the announcements this year, partnerships with leading AI companies is essential— there is potentially a win-win, with these companies looking to expand beyond their traditional markets, train their model on other regional languages such as Kiswahili.
Q: How do we fix the public-private model when it comes to AI innovation?
One issue with the current PPP models is that they’re not built around technology infrastructure as they have been designed roads, bridges, and similar projects. The traditional money institutions that structure financing for technology related investments also require capacity and innovation. It is no surprise that a majority of technology developments in Africa for instance, are largely driven by VCs or external private investment, and not public-private partnerships or local banks. I always joke that if the current big tech companies were from Africa, they wouldn’t exist as they would not have secured any investments for their ventures.
We need to rethink and reimagine public-private partnerships models for evaluation, especially risks. Notions of due diligence, exists etc. This continent, with its business environment and future growth potential, requires patient capital as its fundamentals are just different, compared to other advanced economies.
For instance, a majority of big tech companies in their formative years were willing to take risks, operate without profit for a couple of years, knowing that these ventures will be profitable in future. The current model in Africa for instance, does not have the same courtesy. We need to adopt the same mindset that is fueling those technology ecosystems for developing countries? It is especially important for our economies as it will give the necessary fiscal space to grow the economy by using technologies like AI to remain competitive and ensure shared prosperity.
Q: You’re part of many global conversations spanning various sectors on AI, safety, and security. What are the big gaps in the conversation at the moment?
Inclusivity. When we talk about safety, it’s often within the context of geopolitics. My team has been a part of the International Network of Safety Institutes to collaborate and offer an understanding of what safety, security and inclusivity in our context. The team has also been working on the notion of the need for AI to build public trust. While a majority of discussions are caught in between geopolitics, we have been interested in the opportunities and enablers of AI and its benefits in solving our development challenges from a risk-based approach. We approach these conversations first, from the perspective of inclusivity and trust, then look at safety and security as values that ensures resiliency of these technologies.
Q: Why do you think inclusivity is not part of the conversation?
I don’t think people think about it, and let’s be honest — it’s political. If you talk about inclusivity, it brings in geopolitics. We’ve seen the certain legislations impose restrictions on certain countries to access technology. When I sit in these meetings, I sense that’s the issue. For us, regardless of geopolitics, we’re focused on ensuring Africa is not left behind in this technology. Leapfrogging may not be possible this time because of the serious investments required. This isn’t like moving from landlines to mobile phones. In this case, there has to be a deliberate effort to investing in compute, talent, datasets, including linguistic diversity in training large language models which requires huge investments.
If you think about the four pillars of artificial intelligence—computing, data, talent, and algorithms—computing is about processing capability, which can’t be everywhere. You need trust to leverage compute across borders as workloads are primarily data. In this regard, we have been working on the concept of sovereign AI that will be anchored on trust. While compute and data centers are most likely going to be available in a few countries, we want to ensure that others have the same access, regardless of location, in a safe, secure and sovereign instance. Talent also will and should be shared to work for society’s benefit. Finally, algorithms must be unbiased and engender public trust, ensuring AI is free from manipulation, is explainable, transparent and includes a human in the loop in its decision making.
We must push for multilateralism where sovereignty should not be an excuse for isolationism but should promote global cooperation. We have embassies that are sacred spaces for countries, but it doesn’t mean we’re isolated. We’re working on a white paper to define what sovereignty means in the age of AI that should be about multilateralism and global cooperation for shared prosperity.
Q: Your work spans digital governance, financial inclusion, and technological innovation. Can you share a specific project you have worked on and its impact on Kenya?
I’m excited about our collaboration with IBM and Bin Zayed University that built one of the first AI models for climate action. This model supports H.E President William Ruto’s agenda to increase Kenya’s forest cover. We leveraged this model to create an initiative for Conservation At Scale that brings together multiple actors to invest in afforestation, just energy transition and creates green jobs for young people. The model was further fine tuned to respond to floods in collaboration with the Red Cross, Microsoft, Meta, Google, and Amazon, allowing us to map land slide risks and send alerts to people in potentially risky areas around floods and landslides.
Another project is one improving farmer outcomes and increasing value chains around dairy. We’ve been able to double farmers' income in six months by leveraging simple compute and data. We’re soon hosting a mega fair to showcase how AI has fundamentally changed milk transportation. We’ve also started an insurance scheme for farmers called Maziwa ni school fees, which means milk. Using AI, we’ve identified that farmers’ biggest spending is on health and education, not agriculture. We have been able to leverage on this extra production of milk to convert the income into a facility that pays school fees for children.
We realised that tracking outputs and quickly calculating what is needed to improve output is crucial. This involves multiple factors like feeds, weather, and individual cow health. We’ve embedded the AI model into Meta’s WhatsApp to provide real-time feedback to farmers in PDF. This has built a trust ecosystem, allowing farmers to know that the milk they gave at the farm gate reached the cooperative and matched the recorded figures. It also allows them to have provenance of their produce at their finger tips.
Let me note however, that models and compute are expensive and is a challenge for Small and Medium enterprises as costs are prohibitive to scale. I have used this example to make the case in policy and decision making spaces on the need for Africa to build its own models that are less expensive and more culturally nuanced. I have also been making the case for a differentiated pricing model for accessing LLMs from AI leading companies if our entrepreneurs are to compete fairly in this economy.
Ко може да гради вештачку интелигенцију? | Apolitical
