Against the Odds

Latin America’s Quiet Leadership in the Age of AI

Photo Credit: Andres Pedraza.

Working on Artificial Intelligence (AI) policy in Latin America has been one of the most rewarding and transformative experiences of my professional life. Over the past few years, I’ve had the privilege of advising more than 15 countries in the region on their national AI strategies—each with its own challenges, lessons and moments of real impact.

I’ve learned the value of conviction. You have to believe in the work—even when others don’t, and even when they ask you to stop. I experienced this firsthand in 2019, when I left the Berkman Klein Center at Harvard to serve as an advisor to the Colombian Presidency. It was a major step in my career—one that could have ended quickly, but fortunately lasted three years. Early in that role, I made a bold move: I proposed that we reframe the country’s digital agenda by renaming the national initiative the Digital Transformation and Artificial Intelligence Policy—placing AI at the core of Colombia’s future.

Today, that may sound like an obvious move. But at the time, the idea faced skepticism—even resistance. Many questioned whether AI deserved such prominence. Some important leaders in the government thought the focus should remain broad, while others doubted the practical relevance of AI for Colombia. Some warned that prioritizing AI would be a long-term mistake.

And yet, six years later, no one refers to the initiative by its full name—everyone simply calls it Colombia’s AI policy. The strategy became a foundational milestone for the country in this area, shaping its voice and vision in a world now defined by rapid advances in AI. What once seemed like a risk is now seen as a model. As the saying goes: time proved us right.

This example illustrates a tension that has long defined the region: a desire to lead, coupled with a deep-seated fear of getting it wrong—or of neglecting what is perceived as important in the long term. This sense of immediacy, combined with persistent doubt, is essential to understanding both what has been accomplished and what still remains to be done.

 Overcoming Caution

Still, I’m often struck by how cautiously AI is approached in Latin America—almost with suspicion, as though it were something foreign or far-off, rather than urgent and inherently ours.

And yet, the region continues to advance—sometimes even despite itself. Back in 2019, only a few countries in Latin America and the Caribbean had formal AI strategies. Today, more than twenty have developed national policies or roadmaps. This progress is undeniable. But one concern continues to drive my research—and it sits at the heart of much of my academic work:

Are we building a regional vision for AI that protects our people, reflects our values, and allows us to remain relevant in the global conversation?
And just as crucially: Are we overcoming the hesitation and lack of conviction that too often holds us back in technology policymaking?

Leadership, Momentum and Domino Effects

To understand what’s been achieved, we need to go back to the beginning. When Colombia launched its national AI strategy in 2019, it wasn’t just producing a document—it was lighting a spark. The initiative benefited from the leadership of countries like Argentina, Chile and Uruguay who were moving in parallel. Together, we created a kind of domino effect. Soon, even countries that hadn’t yet considered AI began exploring its implications and opportunities.

What struck me most, however, wasn’t just the momentum—it was the genuine curiosity and drive to learn. Decision makers in these countries asked whether AI could improve educational outcomes. Local leaders wondered whether it could help small farmers adapt to droughts, improve disaster preparedness or make public transportation more efficient. In some of the early meetings, I saw more openness than fear—a rare and beautiful thing in policy circles.

Development banks and multilateral organizations played a key role. They offered more than funding—they provided frameworks, legitimacy and platforms for collaboration. Thanks to their support, AI moved from buzzword to priority on regional agendas. 

From Strategy to Action

Despite the early energy, many strategies have struggled to make the leap from design to implementation. And here’s where my optimism is tempered by experience. I once asked a government official how their national AI strategy was being implemented. He smiled and said, “Well, that was written by the last administration, so it’s not really our priority. But it’s a very well-written document.”

That response—while candid—was also disheartening. I knew how hard the previous team had worked on that document. But they hadn’t built in the political or institutional foundations needed for continuity.

We’ve seen it time and time again: a change in administration sends digital agendas to the back burner. Budget constraints stall pilot programs. A lack of technical talent makes even modest goals difficult to execute. As a result, too many national AI strategies remain more aspirational than actionable.

What concerns me most isn’t just the inefficiency—it’s the opportunity cost. Without long-term investment in public data infrastructure, digital inclusion and ethical oversight, AI risks becoming either a corporate monopoly or a regulatory afterthought. Neither scenario supports the public good.

When Regulation Becomes the Only Conversation

In recent years, I’ve noticed a shift in the AI conversation in the region—one that’s both fascinating and concerning. Many governments are now focused primarily on how to regulate AI, rather than how to use it for public benefit.

To be clear: regulation is essential. The risks posed by AI—bias, surveillance, disinformation—are real and urgent. But in many countries, the conversation has tilted too far in one direction. Regulation is being treated as the starting point, when in fact it should be one of many tools.

I’ve lost track of how many AI-related bills are currently circulating in regional legislatures—but the number is easily more than a hundred. Some are well-crafted. Others seem like reflexive responses to headlines or pressure from abroad. Entire research teams are now dedicated to tracking and categorizing these legislative proposals.

This rush to legislate before fully understanding the technology or its context has consequences. It leads to misaligned rules that can stifle innovation or fail to address real risks. More worryingly, we’re not investing enough in the tools and capacities that make regulation work: independent oversight bodies, data audits, civic engagement, and strong institutions.

We need a better balance—one where governance enables as much as it restrains.

 Five Priorities for the Decade Ahead

So, where do we go from here?

Over the past year, I’ve worked with colleagues across the region to reimagine what a stronger, more adaptive AI agenda could look like—one grounded in the region’s realities, not imported models or abstract fears. Here are five guiding priorities that have emerged:

1. Stay ahead of rapid developments.
   AI is evolving at breakneck speed. New models—like agentic AI, which enables autonomous decision-making—are emerging faster than policymakers can respond. We need regional mechanisms to monitor, understand, and anticipate these changes, rather than react to them after the fact.
2. Harness AI for public problem-solving.
   AI can help address some of Latin America’s most pressing challenges—from biodiversity conservation in the Amazon to improving access to quality education. We must invest in public sector innovation and empower governments, cities, and universities as testing grounds for responsible AI use.
3. Shift the focus from disruption to productivity.
   The narrative around AI and jobs has often been one of fear. But what if we focused on how AI can enhance—not replace—human labor? With the right policies and training, AI can drive inclusive productivity gains rather than displacement.
4. Prepare for the energy cost of AI.
   Advanced AI models consume vast amounts of energy. Are our grids ready? Are our climate strategies accounting for this demand? We must break down the silos between digital and environmental policy before the consequences catch up with us.
5. Learn before legislating.
   Instead of rushing into sweeping laws, let’s invest in regulatory experimentation—sandboxes, pilot programs and iterative policymaking. AI policy may become one of the most consequential regulatory domains of our time. Let’s get it right, even if that means slowing down.

 This Moment Matters

After eight years immersed in regional AI policymaking, one thing is clear: Latin America and the Caribbean can lead—not just follow—in shaping the future of technology.

We have the talent. We have the institutions. What we often lack is belief—belief that we are not too late, that our problems are worth solving, and that we deserve to help shape the rules of this new era of technological change.

Leadership takes more than bold statements. It requires durable institutions, inclusive processes, regional cooperation, and a deep understanding of our own values and realities. It requires balancing innovation and protection—not choosing one over the other.

As the world debates how to govern the rise of advanced AI, we face a choice:
Will we shape our own future—or will it be shaped for us?

The answer won’t be found in strategy documents alone. It lies in the choices we make today—in classrooms, ministries, research labs and communities. And it must reflect a collective effort—not the agenda of a single person, organization, or government.

If we act with care, imagination and ambition, the region probably won’t just catch up. It will quietly lead—and inspire others to follow.