The AI Impact Summit in New Delhi last week felt like two conferences happening in the same city. Inside the Bharat Mandapam convention centre, the mood was expansion: big names, bigger numbers and a deliberate statement that the Global South could host the kind of gathering that up until now had only been seen in the rich world.
Outside, young men were repainting the railings on Mathura Road in peak morning traffic, freshening up the route for arriving Silicon Valley CEOs. The stark juxtaposition was illustrative of the summit’s central question: who, exactly, is AI being built for?
Reportedly, the event saw the formulation of more than $200 billion in AI investment pledges and deals. Prime Minister Narendra Modi pitched India as a global AI hub. UN Secretary-General Antonio Guterres called for a $3 billion fund to help developing countries build AI capacity. India showcased the digital public infrastructure that has brought 1.4 billion people into a functioning digital economy and shown what serious, state-backed deployment looks like.
But while AI is in boom time, the system that has partly financed global development is heading for a cliff. Official development assistance from OECD donors fell 6 per cent in real terms in 2024, and the OECD says there was likely a much greater decline – somewhere between 9 and 17 per cent – in 2025. The US has halved its foreign aid budget and dismantled USAID. The aid sector is haemorrhaging, with no realistic timeline for recovery.
AI cannot replace aid, which is a fundamentally human project involving real people giving food, water, medicine and compassionate care. The people at the sharp end of global aid cuts will not be rescued by a well-deployed language model. The worst version of "AI for development" is the one that doubles as political cover: donors retreat, a tech demo arrives and everyone pretends it balances out.
But the development sector also has to reckon with where it is now and where it might be in another two or three years if the political fight to restore it continues to fail.
This is where AI can help. The technology is already doing things in development contexts that were not possible five years ago. Predictive tools are anticipating disease outbreaks before health systems become overwhelmed. Climate models are being personalised down to the village level, giving smallholder farmers actionable guidance on when to plant and when to harvest. The AIM for Scale programme – a UAE-led initiative announced at Cop28 – delivered AI-enabled monsoon onset forecasts by SMS to 38 million farmers across 13 Indian states last year, and is expanding to 11 countries across Africa, Asia and Latin America in 2026.
That kind of reach – precise, low-cost, deployable through infrastructure that already exists – is exactly what a sector operating on a shrinking budget needs more of.
This efficiency extends to the aid “back office”. Today, development money often flows through layers of expensive international contractors, not always because they deliver better outcomes, but because they ostensibly can shoulder the compliance, reporting and audit requirements designed to prevent fraud and bribery. USAID’s own data shows direct funding to local partners sits at just 12 per cent. AI tools that automate these workflows can make accountability cheaper, without limiting oversight.
To turn this potential into a more widespread reality, three groups need to work together. First, donors, philanthropies and development banks need to treat AI deployment as core infrastructure. If AI is going to help sustain essential services in a world of shrinking budgets, it needs funding the same way clinics are funded – with money for training, data governance, integration and upkeep. The gap is already significant: AI adoption stands at 24.7 per cent in wealthier countries and 14.1 per cent across the Global South, and it is widening. Africa has less than 1 per cent of global data centre capacity for 18 per cent of the world's population. Closing that gap is a precondition for AI and presents a commercial opportunity worth trillions.
Second, AI companies need to build serious product lines for low-resource settings. If this is genuinely meant to be the great equalising technology of our era, “AI for Good” needs to be a business model, not a comms project. The market for tools that work in development and public-sector contexts across the Global South is large, underserved and growing. It is also demanding, having to develop in an environment of inconsistent connectivity, older devices, dozens of languages, high stakes and thin margins. Delhi showed some genuine ambition. Google demonstrated live speech-to-speech translation in more than 70 languages and Anthropic announced it is curating training data in 10 widely spoken Indian languages. These are the right directions. But the critical question is whether these tools actually improve outcomes per dollar of aid spent through efficiency, better targeting, reduced admin or higher-quality deployment. We do not yet have the evaluation frameworks to answer that question rigorously and building them is one of the most important gaps in the entire AI-for-development space. Without that evidence base, we are taking it on faith that the technology helps.
Third, governments across the Global South need the capacity to evaluate and govern the AI tools they adopt. This doesn’t have to mean training and building their own large language models. But it does mean being able to scrutinise the ones they use, including understanding how they make decisions, where they fail and whether they perform for their populations. The New Delhi Frontier AI Commitments launched at the summit last week are a start.
The UAE's announcement that it will host the AI Impact Summit in 2028, following its co-chairing of next year’s edition with Switzerland in Geneva, puts it at the centre of this conversation at the right moment. The UAE leads the world in AI adoption, with 64 per cent of its working-age population using AI tools. It drove the inclusion of AI at Cop28 and played a central role in the UN Sustainable Development Goals frameworks that remain critical to the Global South. The 2028 summit is an opportunity to move past announcements and show AI deployments that have actually delivered for the populations that most need them. Otherwise, the promise that AI “will benefit all of humanity” is just a slogan.
