The rapid growth in data is driving the emergence of countless new applications that aim to solve a variety of problems affecting our society. This is the case with satellite images, which can help detect various environmental phenomena such as fires, water leaks, or floods, among others. If we apply Artificial Intelligence (AI) for machine learning—an area of AI that essentially learns from examples—their applicability has unimaginable potential.
Is the use of these technologies for problem-solving reliable?
In most cases, the answer is yes. However, AI-based models lack a feature that is vital in public matters: explainability. Intelligent models can often function with a certain degree of error, which may be acceptable, but when we do not understand the reasons behind certain decisions, their use becomes unreliable.
Within this framework, depending on the problem one seeks to address, it can be determined whether AI-based models are useful or not. For now, using AI in any situation remains debatable, as it largely depends on the circumstances and conditions under which a particular solution must be fulfilled.
On the other hand, although images are often easy to acquire and in many cases free of charge, in Latin America we generally depend on the availability and access to such inputs provided by private companies from the developed world. This not only limits the possibilities for developing countries to create innovative applications but also keeps the region in a vulnerable situation, as there is no certainty about future access to these data.
Therefore, as long as the region cannot develop its own satellite technology capable of generating quality data accessible to Latin America, we must not forget that for any solution we wish to address through the combination of AI algorithms with solid databases, we will remain dependent on third parties.
Applications that combine AI with satellite imagery
Beyond the real or theoretical limitations, there are already countless applications in the region that are providing solutions to real-world problems. An interesting example of this system’s applicability is the estimation of poverty indices through nighttime images, where the relationship between the amount of artificial light and the economic income level of a given area is analyzed. Another example is the detection of water leaks through satellite images, which makes it possible to estimate the volume of such leaks and their consequences in the area.
A more delicate example of the application of this technology is the use of images for the detection of clandestine graves in Mexico. Some researchers at the Center for Research in Geospatial Information Sciences (CentroGeo) have worked for more than a decade on generating models capable of detecting areas where there is a higher probability of clandestine graves. This application has had a huge impact on groups of searching mothers, among other organizations.
AI- and satellite-based solutions have endless applications, and as technology advances and access to open data becomes easier, progress will continue toward solving new societal problems. But beyond expectations, for now in Latin America we must adapt to current conditions and create sufficient mechanisms or safeguards so that, if our access to these data becomes limited, companies and organizations can continue providing their services.
However, beyond the region’s technological vulnerability, we must not close ourselves off to these new technologies because, more than a threat, they offer an opportunity for growth—to solve problems more quickly and perhaps with less effort. Let us embrace technology in our daily lives and learn to use it responsibly, ethically, and appropriately.
This text is part of the collaboration between the Organization of Ibero-American States for Education, Science and Culture (OEI) and Latinoamérica21 for the dissemination of the platform “Voices of Ibero-American Women.” Learn more and join the platform HERE.
Autor
Doctora en Tecnologías de la Información y sistemas informáticos por la Universidad Rey Juan Carlos, España. Profesora titular en el Centro de Investigación en Ciencias de Información Geoespacial (CentroGEO) de México.
