ECONOMY: DIGITALISATION
Bridging the AI divide: ensuring equitable technological advancement
a digital disadvantage, from a lack of critical energy and water resources to brain drain. But the fact that much of the world is a ‘compute desert’ with limited capacity to develop AI at all has stark implications for governance. Countries hosting AI in data centres and chips have significant advantages in shaping the rules and norms of this technology. If AI capacity remains concentrated, humanity risks creating a world where its benefits – improved health care, enhanced education and better climate monitoring, to name a few – are unevenly distributed. Such an outcome could deepen existing global inequalities, as developing countries struggle to compete in an increasingly AI-driven world. It may be cold comfort, but we have been here before. We know what is at stake. LEARNING FROM THE PAST Twenty-one years ago, another UN secretary-general, Kofi Annan, addressed leaders from over 175 countries at the World Summit on the Information Society in Geneva. Back then, just 12% of the global population had internet access. This mirrors uneven AI capacity with a hugely important difference: developing countries had a seat at the table and a voice to bend the arc of digital progress. WSIS was born out of the imperative to reduce the digital divide separating richer and poorer countries – and to use multistakeholder cooperation as a catalyst for digital development. This allowed us to accomplish
Multistakeholder cooperation will ensure the socio-economic benefits of AI are shared equitably across the world, preventing a widening digital divide
T his summer, United Nations secretary-general António Guterres delivered a historic address to the International Telecommunication Union Council at our Geneva headquarters. He challenged ITU – the UN agency for digital technologies – to help ensure that “AI never stands for advancing inequality”. However, if we do not level the artificial intelligence playing field quickly enough, we risk creating a new digital divide, with a third of humanity still offline and the Sustainable Development Goals in dire need of rescuing. LOCATION MATTERS A new study mapping public cloud ‘compute’, a critical component of AI infrastructure, shows much of the world’s AI development capacity is concentrated in just 30 countries. In fact, only a handful of the top 100 high-performance computing systems able to train large AI models are hosted in a developing country. Of course, multiple factors put developing countries at
Doreen Bogdan-Martin, secretary-general, ITU
116
G20 BRAZIL: THE RIO SUMMIT — 2024
globalgovernanceproject.org
Made with FlippingBook. PDF to flipbook with ease