One certainty about artificial intelligence is the speed with which it has advanced in recent years – a reflection of its increasing relevance in many of our lives.
Yet it would be a mistake to conflate pace with progress if the technology grows unevenly across geographies and within nations, and as long as the guardrails necessary to protect the vulnerable from potential associated threats are insufficient.
It’s no longer enough to ask in what ways AI can benefit humanity, but instead how it can be made to work for everyone. The New Delhi Declaration, endorsed by 86 countries during last week’s India AI Impact Summit, represents a decisive step towards answering this question. Signatories include the US, China and the UAE, which sent a strong delegation to the summit, led by Sheikh Khaled bin Mohamed, Crown Prince of Abu Dhabi.
The premise of the declaration is simple: AI must be shared equitably across humanity, and not concentrated in the hands of a few. It covers seven pillars, including harnessing the technology for social good.
The declaration comes amid some concerns, particularly in the Global South, that a handful of western technology companies are dominating AI development; and that they are building systems trained largely on western data, reflecting western cultural assumptions and serving western economic interests first.
For example, studies have shown that some facial recognition tools perform worse on darker skin tones. Translation algorithms stumble over languages underrepresented in training datasets despite being spoken by hundreds of millions of people. Healthcare diagnostic tools produce less reliable results for certain ethnic groups.
The problem is that once exported, these AI products may create a new form of dependency in markets across the Global South where residents end up becoming consumers of a technology that don't really serve them. Speaking to The National in late 2024, Carme Artigas, co-chair of the UN’s AI High Level Advisory Body, suggested that the exclusion of non-western perspectives from global AI frameworks was a “new way of techno-colonialism”.
While some tech giants are trying to address these biases, there’s more work to be done.
Meanwhile, substantive efforts are being made to resolve these issues in the Global South itself – the most recent examples of which are two new partnerships, both involving the UAE, that were formalised on the sidelines of the New Delhi summit.
Especially noteworthy is a deal between Abu Dhabi’s technology holding group G42 and the US-based Credo AI. Aimed at building accountability into AI systems from the ground up, the project will entail designing risk-monitoring frameworks and education programmes for governments and enterprises.
G42 will also be involved in a joint supercomputer project in India – bringing together Mohamed Bin Zayed University of Artificial Intelligence, the US-based Cerebras Systems and India’s Centre for Development of Advanced Computing – to establish the kind of sovereign AI infrastructure that allows nations to develop and govern their own technological futures rather than simply licensing them from abroad.
India’s decision to partner with the UAE in these projects underscores the latter’s leadership in AI adoption, and its success in integrating the technology in several sectors of its economy. That it is now focused on helping to build AI systems across the Global South that are transparent, fair, secure and accountable – tenets of what is called “Responsible AI” – is an acknowledgement of what is at stake.
Responsible AI is precisely the kind of governance framework that the moment demands. To that extent, the breakthroughs achieved in New Delhi last week could have far-reaching and long-term implications.
