By Gustavo Fonseca Ribeiro*
Every few years, the digital transformation of our times ushers in an emergent piece of technology as the next steppingstone in human progress: The Internet, cloud computing, blockchain, and now artificial intelligence (AI). Some of the praise is grounded in truth, and some of it is hype. Still, each wave presents the developing world with an opportunity for social and economic progress.
The impact of groundbreaking technologies can be framed in two ways. First, their use can solve societal challenges and bring about efficiency gains in productivity. Second, with enough demand, technology brings forth an ecosystem of socioeconomic value that comprises the activities of its upstream supply chain.
This is best illustrated by AI’s momentum currently and the potential of the wide ecosystem that supports it. AI’s ecosystem includes materials, such as natural resources and infrastructure (e.g., data centers, compute); data, including its processes of production (e.g., data mining) and sources (e.g., copyrighted material, census data, Internet of Things); and human labor, such as developers, data annotators, data scientists, and AI researchers. Moreover, labor extends to context- and domain-specific experts needed to develop and maintain products in specific fields (e.g., agriculture, healthcare, finance, language, justice) because AI is a general-purpose technology.
There are many reasons why Latin-American countries should actively explore these opportunities. This includes the chance to generate socioeconomic value and improve quality of life, empower citizens to address their societal problems and self-determine the progress of technology, and to safeguard digital sovereignty.
First, there is permanent socioeconomic value to be created from the ecosystem that supports technology. Indeed, opportunities lie not only on the outcomes of AI use cases. A study by Acemoglu shows that the productivity gains stemming from AI over the next 10 years in the US economy are currently overstated. Parallel to that, each component of the described ecosystem can function as a node for better living standards across AI’s value-chain. However, these nodes can only support quality of life if developed with fairness and due regard for negative externalities, such as fair labor practices (e.g., formal jobs) and green infrastructure.
In fact, Latin America holds comparative advantages on some of the requirements of AI, among which renewable energy is most prominent. The International Energy Agency’s Outlook for Latin America reports that renewables generate 60% of electricity in the region – twice the global average. Moreover, this is foreseen to increase to 80% by 2050. At a time when humanity is quickly depleting its climate allowance, renewables are critical for expanding AI, a technology that is increasingly energy-intensive as it scales.
The rationale of leveraging AI ecosystems – as opposed to only solutions – to improve living standards is shared with other parts of the Global Majority. In November 2023, African thought leaders met in Nairobi for a workshop on the implications of Generative AI for the future of work in Africa. They agreed that “AI in Africa should support the substantial informal sector, emphasizing empowerment, entrepreneurship, and job creation rather than mere efficiency” (p.30). In March 2023, multisectoral specialists from Latin America similarly came together to issue the Montevideo Declaration on Artificial Intelligence, which underscored: “Improving quality of life, working, economic, health, and general wellbeing conditions must be our priority.”
Second, ecosystem-level progress empowers locals to steer the development of technology towards it being fit-for-purpose to their socioeconomic needs. At minimum, it enables adaptation of technology developed in and for foreign settings to one’s own context. For example, increased computational power is directly correlated with better AI capabilities, and access to compute is a hurdle for AI development everywhere. The Oxford Internet Institute shows that only three companies (Amazon, Microsoft, and Google) hold approximately 70% of the global infrastructure for cloud computing. Besides, there is a contrast between its presence in the developed and the developing world, including Latin America.
(Source: TeleGeography, 2024)
AI can only solve Latin-American problems if developed by those who face them. It can best serve Latin-American users if designed pursuant to their socioeconomic needs. Currently, access of local “problem-owners,” such as researchers and developers, to computational power is an unmet precondition for the region to start developing AI that solves Latin-American problems. Thus, the region’s public and private sectors should prioritize expanding access to computational power.
Third, a mature ecosystem is critical not only to advance AI, but also to equip affected communities with self-determination in steering the field’s scientific progress. As expressed by Chijioke Okorie and Vukosi Marivate, the Global South is not a monolithic entity. Even within each country, there is a variety of communities with different interests and views on AI development. In Natural Language Processing (NLP), a subfield of AI that studies the understanding and generation of human language by computers, Okorie and Marivate explain that these communities may include “owners of (…) traditional knowledge, data scientists and AI developers working on data collection, collation, curation, and annotation; linguists (…); and users” that supply language data.
In the same vein, whether in Africa or Latin America, a strong ecosystem empowers those creating technology to embody it with their cultural values. When it comes to AI, a technology that attempts to model reality from training data, this is key to mitigate risks associated with its capabilities. Local communities live the harms felt by their societies. If there is racism, they feel it in their skin. If there is pollution, they feel it in their lungs. This allows them to prevent AI from contributing to or automating existing harms.
Fourth, on a national purview, a built-up ecosystem safeguards digital sovereignty in two ways. One, the design of digital technologies embeds the values of its creators and regulates the behavior of those subject to its effects. This means AI can be designed with the values of one country and regulate the citizens and circumstances of another when deployed elsewhere. Locals can develop AI with data, models and purposes that are in-tune and can be discussed with basis on their cultural values, rather than relying on systems modelled after other societies, such as China, Europe, or North America (p.27). Moreover, locals are equipped with the skills to scrutinize and adjust foreign technology to be aligned with their cultural values.
Additionally, digital sovereignty is a game of calibrating interdependencies vis-à-vis other countries. The many nodes of an AI ecosystem require a balancing act. Some of them can be created nationally, some may be cheaper or better if foreign, but most require a mix of national and foreign (e.g., cloud computing infrastructure). As digital technology becomes increasingly essential to modern life, governments have to strategize a variety of sources to supply the ecosystem that underpins their digital space. Building up the local ecosystem also decreases Latin America’s dependencies on other regions and may even support further integration across a shared goal of leveraging technology for permanent and sustainable socioeconomic transformation.
*Gustavo Fonseca Ribeiro is a Brazilian lawyer with a Master of Public Policy in Digital Technologies from Sciences Po, Specialist Consultant for Digital Transformation and Artificial Intelligence at UNESCO, and Youth Ambassador for the Internet Society. The views herein reflected are the opinion of the author in his personal capacity and do not represent in any way the position of the organizations to which he is related.
