EL TIPO DE CAMBIO DE HOY ES DE: Thirty minutes before taking the stage for his keynote, “Designing AI for impact in global health, principles and techniques” at the IA SUMMIT LATAM 2025 organized by Universidad Galileo and Microsoft, Padmanabhan Anandan, who was the founder and managing director of Microsoft Research India, sat down to talk with República.
His eyes lit up as he moved between memories and predictions. He recalled how, 45 years ago, tasks that today feel routine—like identifying a face or interpreting an X-ray—were almost philosophical challenges for computer science.
With a calm, reflective tone, he explained that AI not only solved those puzzles but also opened an entirely new frontier for countries with limited human resources. And while he recognizes that the technology will reshape work and daily routines, he insists it ultimately expands human capability.
What sparked your curiosity about teaching machines to “understand” what they see?
—I started working on this 45 years ago, when I began my PhD, and many of the problems we were facing at that time were extremely difficult. For example, recognizing faces, identifying objects, and signs. Those tasks seemed almost impossible. Now they’re easy, not surprising, but back then they were among the hardest challenges in the field.
Another area I worked on was analyzing films and videos. We made major advances in the 1980s and 1990s, and what I see today is that many of the ideas remain, though implemented differently.
It’s nice to know that some things don’t change, but they’re better now because we have more powerful computers. Processes that used to take an entire day now run in seconds.
I think one of the areas where I’ve contributed the most is medical image analysis. Many issues —like reading X-rays or MRIs or detecting difficult conditions— can be solved by machines far more efficiently than by any human today. We can train them on complex cases, and that will make a tremendous difference in the future.
Do you remember the moment you decided to study artificial intelligence?
—Yes, in 1980. I was working as an engineer. It was fun, but not satisfying. I wanted to answer a deep scientific question. And I felt the most important question, the one for which we had no answer, was: “How do we think? What does thinking mean?”
That curiosity led me to pursue a PhD. We understood physics and biology well at the time, and we knew DNA; but when it came to the mind, we knew very little. Interestingly, today I see much more connection between how we think about physics and how we think about the mind.
What is one simple everyday example of how AI can make life easier or safer in countries like Guatemala?
—Mainly in health and medicine. A gigantic portion of the world’s population —perhaps half— does not have good access to medical care, simply because there aren’t enough doctors.
AI can support people who are not highly trained doctors but who understand the basics of providing care. It expands access to services that were previously impossible. It allows doctors to support communities they would never reach on their own.
But every profession will need AI to make work easier. Especially in sectors where there aren’t enough experts, like agriculture. And countries like Guatemala rely heavily on that industry.
What would you say to people who think AI will take their jobs?
—Yes, it will replace some jobs. But I’m not entirely sure how everything will play out, because this is more complex than previous technological shifts. Technology has always taken over certain tasks because it makes them easier, which reduces the need for personnel. However, people have always found new jobs because technology creates new opportunities.
Technology helps us focus on what really matters. Humans can concentrate on the critical parts of their job and leave the mundane aside. I think that will be a significant benefit.
As for whether AI will take all jobs, I really don’t know. It’s possible. But we’ve always found ways to work with technology for our own benefit. I hope we’ll do the same again.
What experience convinced you that AI can truly help low-resource communities?
—I’ll give you two examples. The first is in health. When a newborn arrives, it’s crucial to measure the baby’s weight accurately. If the baby weighs less than 2.5 kilograms, they must be taken to the hospital; if it’s below 1.8 kilograms, to the emergency room. But weighing newborns has not been done well because many people use simple spring balances.
Now, AI can estimate weight from just a few photos of the baby. That can dramatically improve care.
The second example is agriculture and detecting pest infestations before pests begin eating the crops. Once they start eating, you don’t have much time. But with AI, we can predict early on, based on conditions or insect prevalence, whether an infestation is likely.
What is one realistic way AI could help Guatemala without requiring major public investment?
—In many clinics, long lines of people are waiting. And on the front-line work people who are not doctors; they often visit villages and homes. Many don’t even have a high school education.
Rather than trying to redesign the system, AI can help them do their jobs better. It can enable diagnoses at home, without requiring a trip to the clinic. Many tests that doctors perform today could be done using AI-supported kits. The person doesn’t need to understand the technical details, the machines can read the results, though a human is needed to administer the test.
What have doctors and communities taught you about designing technology people can trust?
—Interestingly, poor communities and frontline workers trust technology more than you might expect, simply because they don’t have many alternatives. What matters to them is that the system addresses their real problems: nutritional issues, available medications, or treatments allowed for cultural reasons. Often, technologists don’t understand these contextual details.
We’ve learned that they need and request solutions designed around their realities.
What lessons from India’s experience in technology and education could inspire Guatemala?
—India is a country that is both developed and developing at the same time. It has technology infrastructure and the capacity to produce highly qualified talent, but it also faces deep poverty.
India’s biggest long-term investment has been in professional education, particularly engineering and science. That investment was made 70 years ago, when the population was young. Today we’re seeing the benefits. And that investment had to come from the government through the public education system.
How should leaders decide whether adopting AI is the right move for their country?
—The same way they evaluate any technology. When leaders encounter an innovation, cell phones in the past or automobiles before that, they determine whether it should be adopted, how it should be applied, and its economic impact.
New technologies create new opportunities, which means new winners and losers and potentially corruption. There’s no difference now. Whatever countries have done to manage other technologies is what they should do for AI.
How would you explain to a teenager what AI is and what it isn’t?
—I’d say it’s many things. One of the most valuable human abilities is reasoning about situations: seeing something happen, understanding why, what it means, and what will happen if we act. That process is complex.
What we’re doing now is building machines that can do the same and understand why they see something, how it works, and what to do. We can get machines to think like we do and then talk to us. We’ll have another set of “friends” who can help us solve problems.
How can AI help people living in countries with low technological adoption?
—Health is key. That’s where AI can make the most significant difference, because the problems are so hard. It simplifies diagnosis, access, and treatment, and democratizing healthcare is truly possible.
AI can also create opportunities for businesses and small enterprises. It allows them to manage their operations without relying on intermediaries, which is empowering.
Of course, there are risks and the technology is complex and requires significant computing power. But on the usage side, it can empower many people.
What message do you want people to remember after your talk?
—To achieve real social impact with AI, you need collaboration between policymakers, technology providers, NGOs, and donors. Without that collaboration, nothing is possible.
Thirty minutes before taking the stage for his keynote, “Designing AI for impact in global health, principles and techniques” at the IA SUMMIT LATAM 2025 organized by Universidad Galileo and Microsoft, Padmanabhan Anandan, who was the founder and managing director of Microsoft Research India, sat down to talk with República.
His eyes lit up as he moved between memories and predictions. He recalled how, 45 years ago, tasks that today feel routine—like identifying a face or interpreting an X-ray—were almost philosophical challenges for computer science.
With a calm, reflective tone, he explained that AI not only solved those puzzles but also opened an entirely new frontier for countries with limited human resources. And while he recognizes that the technology will reshape work and daily routines, he insists it ultimately expands human capability.
What sparked your curiosity about teaching machines to “understand” what they see?
—I started working on this 45 years ago, when I began my PhD, and many of the problems we were facing at that time were extremely difficult. For example, recognizing faces, identifying objects, and signs. Those tasks seemed almost impossible. Now they’re easy, not surprising, but back then they were among the hardest challenges in the field.
Another area I worked on was analyzing films and videos. We made major advances in the 1980s and 1990s, and what I see today is that many of the ideas remain, though implemented differently.
It’s nice to know that some things don’t change, but they’re better now because we have more powerful computers. Processes that used to take an entire day now run in seconds.
I think one of the areas where I’ve contributed the most is medical image analysis. Many issues —like reading X-rays or MRIs or detecting difficult conditions— can be solved by machines far more efficiently than by any human today. We can train them on complex cases, and that will make a tremendous difference in the future.
Do you remember the moment you decided to study artificial intelligence?
—Yes, in 1980. I was working as an engineer. It was fun, but not satisfying. I wanted to answer a deep scientific question. And I felt the most important question, the one for which we had no answer, was: “How do we think? What does thinking mean?”
That curiosity led me to pursue a PhD. We understood physics and biology well at the time, and we knew DNA; but when it came to the mind, we knew very little. Interestingly, today I see much more connection between how we think about physics and how we think about the mind.
What is one simple everyday example of how AI can make life easier or safer in countries like Guatemala?
—Mainly in health and medicine. A gigantic portion of the world’s population —perhaps half— does not have good access to medical care, simply because there aren’t enough doctors.
AI can support people who are not highly trained doctors but who understand the basics of providing care. It expands access to services that were previously impossible. It allows doctors to support communities they would never reach on their own.
But every profession will need AI to make work easier. Especially in sectors where there aren’t enough experts, like agriculture. And countries like Guatemala rely heavily on that industry.
What would you say to people who think AI will take their jobs?
—Yes, it will replace some jobs. But I’m not entirely sure how everything will play out, because this is more complex than previous technological shifts. Technology has always taken over certain tasks because it makes them easier, which reduces the need for personnel. However, people have always found new jobs because technology creates new opportunities.
Technology helps us focus on what really matters. Humans can concentrate on the critical parts of their job and leave the mundane aside. I think that will be a significant benefit.
As for whether AI will take all jobs, I really don’t know. It’s possible. But we’ve always found ways to work with technology for our own benefit. I hope we’ll do the same again.
What experience convinced you that AI can truly help low-resource communities?
—I’ll give you two examples. The first is in health. When a newborn arrives, it’s crucial to measure the baby’s weight accurately. If the baby weighs less than 2.5 kilograms, they must be taken to the hospital; if it’s below 1.8 kilograms, to the emergency room. But weighing newborns has not been done well because many people use simple spring balances.
Now, AI can estimate weight from just a few photos of the baby. That can dramatically improve care.
The second example is agriculture and detecting pest infestations before pests begin eating the crops. Once they start eating, you don’t have much time. But with AI, we can predict early on, based on conditions or insect prevalence, whether an infestation is likely.
What is one realistic way AI could help Guatemala without requiring major public investment?
—In many clinics, long lines of people are waiting. And on the front-line work people who are not doctors; they often visit villages and homes. Many don’t even have a high school education.
Rather than trying to redesign the system, AI can help them do their jobs better. It can enable diagnoses at home, without requiring a trip to the clinic. Many tests that doctors perform today could be done using AI-supported kits. The person doesn’t need to understand the technical details, the machines can read the results, though a human is needed to administer the test.
What have doctors and communities taught you about designing technology people can trust?
—Interestingly, poor communities and frontline workers trust technology more than you might expect, simply because they don’t have many alternatives. What matters to them is that the system addresses their real problems: nutritional issues, available medications, or treatments allowed for cultural reasons. Often, technologists don’t understand these contextual details.
We’ve learned that they need and request solutions designed around their realities.
What lessons from India’s experience in technology and education could inspire Guatemala?
—India is a country that is both developed and developing at the same time. It has technology infrastructure and the capacity to produce highly qualified talent, but it also faces deep poverty.
India’s biggest long-term investment has been in professional education, particularly engineering and science. That investment was made 70 years ago, when the population was young. Today we’re seeing the benefits. And that investment had to come from the government through the public education system.
How should leaders decide whether adopting AI is the right move for their country?
—The same way they evaluate any technology. When leaders encounter an innovation, cell phones in the past or automobiles before that, they determine whether it should be adopted, how it should be applied, and its economic impact.
New technologies create new opportunities, which means new winners and losers and potentially corruption. There’s no difference now. Whatever countries have done to manage other technologies is what they should do for AI.
How would you explain to a teenager what AI is and what it isn’t?
—I’d say it’s many things. One of the most valuable human abilities is reasoning about situations: seeing something happen, understanding why, what it means, and what will happen if we act. That process is complex.
What we’re doing now is building machines that can do the same and understand why they see something, how it works, and what to do. We can get machines to think like we do and then talk to us. We’ll have another set of “friends” who can help us solve problems.
How can AI help people living in countries with low technological adoption?
—Health is key. That’s where AI can make the most significant difference, because the problems are so hard. It simplifies diagnosis, access, and treatment, and democratizing healthcare is truly possible.
AI can also create opportunities for businesses and small enterprises. It allows them to manage their operations without relying on intermediaries, which is empowering.
Of course, there are risks and the technology is complex and requires significant computing power. But on the usage side, it can empower many people.
What message do you want people to remember after your talk?
—To achieve real social impact with AI, you need collaboration between policymakers, technology providers, NGOs, and donors. Without that collaboration, nothing is possible.
