The Seoul AI summit should put local capacity at the heart of AI for international development

For the UK, backing the use of emerging technology for development goals should extend to support of local leadership and stewardship.

Expert comment Published 22 May 2024 3 minute READ

This week’s AI Seoul summit offers an opportunity to broaden the conversation on AI from a narrow but important focus on safety, to one that further explores the technology’s potential benefits.

The Seoul summit convenes governments and select global industry, academic and civil society leaders, not only around efforts to ‘ensure AI model safety’ but to support ‘innovation and inclusivity’.

The UK is a co-host of the summit, and hosted the first AI Safety Summit in November 2023, consistently championing AI safety. But it has coupled that with a focus on how AI can bring benefits to poorer countries – announcing an £80 million AI for Development programme at the 2023 summit. The Seoul summit is an opportunity for the UK to reaffirm that focus.

Routes to digital development – the infrastructural approach

The summit comes just two months after the launch of the UK’s Digital Development Strategy 2024–30, as well as the UN AI Resolution and the G7 ministerial declaration on digital innovation. All express hope in AI’s potential to help achieve development goals.

But underneath these shared expressions of hope, states have pursued very different strategies for digital development in practice.

China, for example, has focused on investment in digital infrastructure in developing countries – via its ‘Digital Silk Road’ programme. Its approach offers recipient nations low-cost ‘hardware’ such as 5G fibre optic cables, antennas, satellites, and cloud computing, often via support to Chinese companies like Huawei and ZTE.

Since its introduction in 2015, the exact value of Digital Silk Road investments has been opaque. 

But some findings suggest that, as of 2017, China had provided more financing for tech and digital than all multilateral agencies and leading democracies combined.

Digital infrastructure may contribute to increased productivity and widen access to new technologies. But there are also concerns China’s approach increases its influence over recipient states’ digital governance and data security – and risks dependency on foreign state-owned or state-aligned businesses. This ties the recipient state to the funder state, in a loop of digital reliance and reinforcement.

UK leans towards a market-led approach

While China focuses on supporting the export of its own technology, and on hardware and infrastructure, other digital development approaches concentrate more on shaping markets. 

DFID worked with Vodafone in the early 2000s to incubate ideas about providing mobile money transfer services to poor people in Kenya. 

These approaches tend to focus on developing digital services, skills, and regulations that can enable economic growth and include poor people in markets and finance systems. They are often delivered via partnerships with the private sector, including global tech firms.

The UK has leaned more into this approach – with some success. DFID worked with Vodafone in the early 2000s to incubate ideas about providing mobile money transfer services to poor people in Kenya. 

These ideas eventually became M-Pesa, Africa’s most successful mobile money service. A 2016 study found M-Pesa had lifted 2 per cent of Kenyan households out of extreme poverty between 2008 and 2016.

The UK’s new Digital Development Strategy builds on this approach,emphasizing it will deliver outcomes through partnerships between partner governments, civil society, and the private sector. It references strategic relationships with mobile telecomms companies and other technology firms.

While utilizing corporate expertise brings clear benefits, technology companies are not apolitical. Best practice and good governance must be at the heart of this kind of public–private development to avoid situations where critical services or markets are too dependent on private sector tech entities.

For example, in 2020, the Rwandan government announced a decade-long partnership with a digital health company called Babylon to deliver many of its primary health services. 

But Babylon, a UK-based company which also received support from the Gates Foundation, got into financial difficulties in 2023, undermining the delivery of some of these basic services for many Rwandans.

These examples underscore that the influence and role of tech companies in development should be carefully balanced.

The power of community-driven AI  

Managing the influence of the private sector should also be a priority when it comes to AI for development. Both ‘hardware’ and ‘software’-led approaches to digital development can risk bolstering the role and importance of major tech firms.

In Africa, Asia and Latin America, many AI experts are already doing work that challenges dominant tech and political powers.

Enabling development recipients to strengthen their own AI capacity would provide an important counterbalance: the IMF has underscored the importance of local innovation, adoption and stewardship.

In Africa, Asia and Latin America, many AI experts are already doing work that challenges dominant tech and political powers. 

Lelapa AI is an AI lab that builds AI solutions in Africa which are context-specific to the continent and use local tech talent. The Mandla App utilizes AI to teach languages like Zulu, Xhosa, Somali and Amharic that are poorly served by tools largely trained on US English. The open-source Ghana NLP initiative supports better data sources for training AI tools in West Africa.  

It would be unfair to characterize the UK’s Digital Development Strategy as fixated on large private sector companies as the silver bullet of digital development. The strategy emphasizes the importance of building local capacity and digital entrepreneurialism, through the FCDO’s UK Tech Hubs programme, African Technology and Innovation Partnership and Digital Access Programme. 

It also has the positive goal of creating or scaling up responsible AI research labs at African universities. And it emphaszses that it will seek to work with civil society and local organizations as well as private sector companies.  

From consultation to ‘digital self-determination’  

Involving these stakeholders is good, but it’s important to acknowledge the vast difference in power between the players. 

The UK approach could do more to acknowledge AI is not a monolithic ‘tool’. 

The UK’s – and other donors’ – approach could be strengthened by moving from a focus on consulting local players to a focus on ‘digital self-determination’ – giving people greater control and agency over how technology, data and services are developed and used. 

Article 2nd half

Specifically, the UK approach could do more to acknowledge AI is not a monolithic ‘tool’. 

The current strategy is broad and far-reaching. Building data centres, or rolling out AI education programmes, or incentivizing AI start-ups, are all very distinct applications of AI in development that require different resources, networks and management. 

A tailored approach that allows for partner countries to specialize in a specific AI area could have greater impact.

Additionally, tech companies have been accused of using cheap labour in developing countries to trawl through and moderate offensive material to develop ‘safer’ AI tools in line with long-standing industry approaches to content moderation hubs. The UK strategy could include more on addressing the exploitation of workers in the development of AI.

The UK government has an opportunity with this summit to explore better ways to broker the complex power structures of AI in society: its development strategy and its approach to international convenings should reflect this.