In the first roundtable organized by Insight-LAC, a clear consensus emerged: artificial intelligence is a structural disruption. Its impact spans multiple dimensions—productive, institutional, educational, and environmental—and raises urgent ethical dilemmas. But it also opens a window of opportunity to reimagine the region’s development path.
With the participation of representatives from international organizations, the public and private sectors, academia, NGOs, and the media, the discussion raised key questions: Is the region prepared for the technological shift brought by AI? How can we prevent this new revolution from deepening existing inequalities? What productive and social opportunities could it unlock if strategically planned? And what do we need to do—from public policy, education, research, and governance—to avoid being left behind?
Some of the main points of consensus that emerged from the dialogue:
Environment and sustainability: the other frontier. Artificial intelligence can be a powerful tool to address climate change: it can help anticipate extreme events, optimize the use of natural resources, and even value key ecosystem services for conservation. At the same time, its development and intensive use carry a high energy cost, especially in the storage and training of large models. This double-edged nature requires responsible action. Integrating AI into institutional frameworks with data governance, territorial coordination, and environmental justice is essential. It is also necessary to apply a cost–benefit logic to determine when and how these technologies should be used sustainably. Not all technological solutions are neutral; their impact depends on how and for what purpose they are used.
Current AI is powerful, but still limited. While some speak of a possible “new industrial revolution,” we are still far from artificial general intelligence. What exists today are narrow systems, highly effective at specific tasks but dependent on massive datasets, supervised training, and human-designed algorithms. They still lack autonomy, deep understanding, and contextual awareness. Therefore, we must maintain a critical and clear-eyed perspective: powerful, yes. Magical, no.
Uneven and disordered adoption. There is a coexistence of self-taught adoption by workers and uneven organizational strategies, often driven more by “trend effects” than by a strategic vision. In both cases, there is a lack of comprehensive understanding of the risks, costs, and social and environmental impacts of AI adoption. This reinforces—and deepens—existing gaps between individuals (age, gender, education, social background, or digital experimentation capacity), companies (large firms better positioned, though some agile startups also stand out), and territories (access to infrastructure, data culture, and available talent).
Public policy has a key role. At the state level, the picture is similar. There is interest, there are regulatory discussions, there are plans on paper—but still a lack of robust or specialized institutions, strong governance frameworks, and budget allocations capable of guiding this agenda strategically toward development.
More than productivity: meaning. The question of what we ask algorithms to optimize is deeply political. Productivity is not only efficiency: it is how we combine labor, capital, data, and knowledge to generate value. And that value must align with the kind of society we want to build. AI can help address urgent challenges: improving educational outcomes, diversifying productive structures, anticipating climate impacts, and reducing persistent social inequalities.
Education and work in transformation. The education system faces the challenge of developing new skills without having solved structural deficits. Meanwhile, the labor market must rethink how to support worker reskilling and promote AI that complements—rather than replaces—human capabilities.
Science, technology, and innovation with a regional identity. Local capabilities must be made visible and linked to the productive sector. The region cannot limit itself to consuming technology: it must generate knowledge, train models with its own data, and define its own ethical standards. However, investment in research and development (R&D) in Latin America and the Caribbean remains very low (0.6% of GDP), accounting for only 2.4% of global investment. If we want AI to enhance development rather than reproduce historical dependencies, we need a scientific and technological agenda made in, by, and for Latin America—and that requires investing more, coordinating better, and committing decisively to local capabilities.
Without algorithmic justice, there is no inclusion. AI amplifies the biases present in the data it is trained on. Including gender, diversity, and territorial perspectives is not just good practice: it is a condition of legitimacy. The issue is not only correcting technical errors; AI is also shaping subjectivities—how identities are represented, how decisions are made, and which voices are amplified or excluded. It is therefore urgent to make historically underrepresented groups visible through data and build models that reflect the plurality of our societies. Because if data invisibilizes, algorithms exclude.
In the exchange, a shared concern became clear: after so many debates about the risks and promises of AI, it is time to focus on the “how”. Today, we need to build concrete tools, methodologies, and capabilities that allow AI to be integrated safely, ethically, and effectively into real-world processes—productive, institutional, and educational.
The challenge lies in translating concerns and assessments into operational solutions based on real cases, specific problems, and existing knowledge. This is also the commitment of Insight-LAC: to support with evidence, but above all with tools that help govern and leverage AI in concrete contexts.
This was the first in a series of meetings organized by Insight-LAC aimed at opening urgent conversations about the region’s development challenges.