Generative AI: What Should Governments in Latin America Do?

Para leer en español, haga clic aquí.

Original (English) version on Medium.

Generative AI: What Should Governments in Latin America Do?

By Elisabeth Sylvan and Armando Guio Español

Given ongoing concerns and discussions in Latin America about both labor and digital literacy, fear about how new generative AI tools will impact the region is both understandable and reasonable. Latin America’s unemployment rate is high, seven percent on average. The OECD has determined that more than 25% of jobs in Latin America may be at risk of replacement from automation, one of the highest rates in the world. El Tiempo, a popular newspaper in Colombia, recently created a special edition with which people can self-assess the risk of their employment being replaced soon.

In October 2023, Latin American governments will reunite in Santiago de Chile to discuss regional measures regarding AI ethics and governance, including the emerging regulatory landscape for the deployment of generative AI systems and tools. This meeting will be organized by UNESCO and the Corporacion Andina de Fomento (CAF), a Latin-America-based development bank, as a regional initiative to determine the future of AI governance in the region and to foster collaboration among different countries and organizations. The outcomes of this meeting will be highly relevant not just to policymakers charged with generating and enforcing emerging technology governance for countries such as Mexico, Brazil, Chile, Uruguay, and the Dominican Republic, but also to all those invested in jurisdictions and regulations that apply to the future of work, education, and regional cooperation in the Global South.

Any Latin American civic response will need to be proportional to the real risks to the region and not be reactive to hype and fear. Staunch measures, such as banning innovation entirely, will restrict the region’s ability to impact how new AI tools are designed and deployed. On the other hand, inaction may deepen existing problems, weakening competitiveness and wasting precious time in response to fast-moving technological innovations.

We propose six specific measures that can help governments and decision-makers in Latin America to develop a clear roadmap for navigating this complex environment:

1. Continue with ongoing AI policy implementation while increasing public participation therein

All countries have an opportunity to share information about and increase public participation in AI policy implementation, regardless of where they are in their process. Countries that have designed AI national strategies and those that are considering updating their AI national strategies can increase stakeholder involvement. When people are included, they may be more likely to view their governments as responsive. Inclusive practices will have the added benefit of educating the public about both generative AI and policy decisions.

2. Identify and share the benefits of generative AI to the public good

In Latin America, the narrative about generative AI has been, on average, negative, typically focusing on how tools will replace humans. Governments can work towards a more balanced dialogue by thoughtfully weighing the benefits against the risks and highlighting benefits to the public in an even-handed and fair manner. Generative AI provides governments with a real opportunity to improve the public services they offer through digital government and digital transformation programs. For example, generative AI tools could help to reduce agency response times and increase information access.

3. Define societal guardrails

While the benefits of generative AI are great, so are the risks. Together, we need to define how new tools will be used and how to handle issues such as plagiarism and falsehoods. Countries will need to consider how they will address risks against other fundamental rights. Ethical discussions must be a priority and different models must be piloted to determine how basic principles may be secured. The US’s Blueprint for an AI Bill of Rights provides a good reference on how public entities can operationalize the principles that promote the responsible use of AI.

Governments in Latin America should also promote ethical principles in the private sector. Companies and AI entrepreneurs who are developing or using AI tools must engage with governments on ethical design. In parallel, AI tools need to be vetted. Governments can explore how to effectively incentivize companies, such as ethical certifications for AI applications and how to develop oversight mechanisms at all stages of design and deployment. Stakeholders, especially consumers and civil society, need to be actively engaged in discussions about ethical concerns and encouraged to demand ethical behavior. Thus, governments need to promote the participation of both the private and public sectors while consumer groups should have an active role in providing actions for consumers to take.

4. Accelerate regulatory experimentation and develop a first draft of AI legislation for the region

Before enacting extensive regulations, Latin American countries can test controlled regulatory approaches. Ambitious policy prototyping projects will allow experimentation with innovative regulatory proposals before they are adopted by governments. Regulatory sandboxes and policy prototyping initiatives, for example, will help countries to measure the impact of proposals. Regulatory sandboxes are legal spaces, often temporary, in which companies aren’t subjected to all regulations. These approaches allow all stakeholders to test different models while simultaneously examining how our legal, technical, and social systems will respond.

A key next step is to develop the first draft of generative AI system regulation that will support sandboxes and other prototyping. Just as Europe published a white paper on AI regulation, so should Latin-American countries begin to formalize their regulatory approaches. Though Latin American countries can examine existing models of legislation from elsewhere in the world, their legislation should suit the needs of the region. Countries in the region may wish, for example, to center human rights issues more than technical ones (as the US’s Blueprint for an AI Bill of Rights has done) and on procedure rather than content. Using existing international and regional principles of human rights, governments can test procedures and actions that would implement those principles in AI systems.

5. Increase AI skills across populations, especially among women and children

One growing concern for those considering education and learning ecosystems in Latin America is whether people will have the skills to thrive in a world redefined by generative AI. People have the right to understand generative AI’s influence on their choices, behaviors, and actions and to gain and apply AI-related skills. Government should prioritize educational initiatives not only to support employment but also to support people to think critically about the responsible use of generative AI. The good news is that even though generative AI is new, governments have available existing educational resources. For instance, Finland’s ‘1 percent’ demonstrates that education can both prepare people for employment and reduce their fears. Colombia’s platform, Aprendeia.Org, created through a recent AI expert mission, is a set of free Latin American educational resources that anyone can use. And Youth and Media’s evolving collection of AI tools empowers teachers and learners across geographies and communities. All three approaches can be adapted throughout Latin America.

Though all people have a right to meaningfully engage with generative AI, not all people are positioned well to do so. Significant digital divides persist. Women in Latin America have less access to technology training and have fewer technology-related jobs than men. Unless governmental efforts address these pipeline concerns, women will not be able to contribute to the ethical implementation of generative AI as much as their male counterparts. Also, educational programs need to be developed for young people, both in and outside of school since their lives will be impacted by generative AI over the long term.

6. Make strategic alliances; collaborate rather than compete

Governments in the region do not need to work alone. For example, if countries share the outcomes of their regulation experiments, AI literacy programs, and community engagement approaches, they can learn from one another what will succeed within the region. This collaboration can help countries to avoid repeating one another’s mistakes, saving precious time and resources. And if the region were to build shared data infrastructure for exchanging knowledge, together, Latin America overall would be more globally competitive. Such an infrastructure will facilitate efficient collaboration and strengthen the regulatory, educational, and job development initiatives in the region as a whole.

Governments can also work with external organizations, including the region’s many academic institutions, as well as international organizations that contribute to Latin American development. The exchange of ideas and experiences across countries and institutions will have a synergistic effect on understanding and possibilities. Government can work to generate effective spaces that facilitate alliances and set achievable regular goalposts for incremental outputs.

Now is the time to consider an AI Alliance in Latin America, bringing governments and other organizations to develop a shared agenda.