We need global coordination on data, not just AI; here’s why

Data is foundational to a digital world. Advances in artificial intelligence (AI) over the past year have highlighted its value by enabling more human-like interaction through vast data processing.

Data is one of the three pillars of generative AI, along with computational power and access to talent. It is increasingly the lifeblood of participation in public and private life. More complex and integrated technologies mean data will continue to be more important. A business’s economic advantage is now linked to intensive data use. And as governments undergo digital transformations, data mediates citizen inclusion and participation, from access to public services to identification at borders.

Despite the transforming potential of data, there remains a huge inequity regarding who benefits from its expanding uses, particularly between the majority and minority worlds. While the former has significant data on health, education, economics and social media, they can not draw value from it.

While data equity is being considered globally – such as via the International Monetary Fund’s initiative to address policy-level data gaps in Group of 20 (G20) nations – these efforts do not address fundamental power inequalities to accessing data at the country and firm level or the resources required to derive value from it.

Despite the foundational nature of data and such evident inequalities, the focus of current discourses on digital questions is scattered.

First, there is a global conversation on the digital divide, within which is a sub-discussion on digital public infrastructure (DPI) and access to technology. There is a focus on responsible, accessible DPI that enables people to transact more meaningfully online, as pushed by ID for Africa or another pan-African movement 50in5. The Indian and Brazilian G20 presidencies have promoted DPI as well. However, the DPI focus has not adequately acknowledged the vast data generated through identity, payments or data exchanges.

The second core area of global attention is AI, promoted through now ubiquitous AI summits e.g. recent events include the AI Safety Summit in the United Kingdom on risks of AI development and in Seoul, South Korea, on AI Safety, Security and Innovation. There have also been multilateral forums such as that organized on the sidelines of the World Summit of the Information Society +20 – the AI for Good Summit. The World Economic Forum had its own AI summit too, alongside several smaller ones.

From one view, inequalities in the benefits from data are captured within both areas of global concern. With DPI, for example, global entities such as Co-Develop seek to address global inequalities, in this instance, with a mandate to fund the creation of digital infrastructure in countries. Around AI, the Group of Seven’s AI Hub aims to support the developing world. The process of developing the Global Digital Compact has also given rise to debates over equitable and accessible data and data governance.

So why is data still not part of the discourse on the digital divide, digital public infrastructure, or AI and innovation?

The issue lies in “how” data is considered and to what extent references to data allow policy actors to understand and make decisions about it, given its fundamental and evolving role in all areas of life.

Existing global conversations focus on either (i) the creation of infrastructures that produce data, so do not adequately study the governance of data, or (ii) are concerned with the consequences of the use of data through AI models, so do not meaningfully consider data itself as the input to these applications. As a core input and product of changing digital systems and infrastructures, data isn’t being considered in its own right.

There could be practical reasons for this gap. It can be simply easier to conceptualize and govern instants of data use. AI models and outputs can be easier to view, while “data” in and of itself is tricky to grapple with – is it an economic good or individual asset, or is it relational and collective?

The language used to talk about data and data governance has also struggled to capture the realities of data. The idea of data ownership can fail to capture how data is replicable, produced through multiple actors, and becomes valuable in the aggregate or combination. Efforts that help people have agency over data, such as rights of portability, take place at an individual level, rendering them meaningless beyond specific uses.