Verónica Álvarez, mathematician: “Algorithms are tools that help us make decisions, not replace us.”

Algorithms are chameleons. Just like these reptiles, they camouflage themselves. They adjust their shapes to adapt to millions of users and survive . Fields such as bank fraud detection or the recommendation systems of Amazon and Netflix demonstrate their constant evolution. Verónica Álvarez knows this well. A mathematician by training, born in Zaragoza 27 years ago, she has made this capacity for adaptation her main line of research. Her work was recently recognized with the Young Researchers Award granted by the BBVA Foundation in the field of computer science. And not by chance. Her career is the result of years of studying data and statistical models.

“Consumption patterns change and user tastes vary over time, forcing algorithms to constantly readjust,” he explains from his office in Washington, D.C., in a video call with EL PAÍS. He is temporarily based there while conducting his postdoctoral research at the Massachusetts Institute of Technology (MIT). His goal is to design artificial intelligence (AI) algorithms that not only work today, but also learn to evolve over time. That don't stay still. And that, like chameleons, know how to adapt to the world around them.

Because, after all, algorithms—those sets of instructions that AI uses to process information and make decisions—are only as good as their ability to change with the world. And that idea isn't new to Álvarez. Before joining MIT and until 2024, she was a researcher at the Basque Center for Applied Mathematics in Bilbao. There, along with her colleagues Santiago Mazuelas and José Antonio Lozano, they created an electricity demand prediction tool. A system that also learns from changes in usage habits, applying the same chameleon-like logic that guides her research today: adapting so as not to be left behind.

"Ultimately, in AI, you use different mathematical techniques," he maintains. This breakthrough was based on an algorithm that not only serves to anticipate power grid surges . It could also have other applications, such as in cybersecurity and medicine.

Question : Could you explain how statistics, artificial intelligence, and mathematics relate to your daily work?

Answer : AI relies on mathematics and statistics constantly. We first identify a problem, for example, how to adapt to changes over time. We observe how these changes occur in real life, for example, how what happened yesterday has a greater impact than what happened weeks ago. From this, we use statistical tools to model this behavior, ensuring that the algorithms adapt to observable patterns and human behavior.

Algorithms are tools that help us make decisions or facilitate our work, not replace us.

Q. Do algorithms try to replicate human behavior?

A. That's right. Humans learn over time. We use our knowledge to learn new things. If you want series on streaming platforms to adapt to your current tastes, algorithms use your past decisions to recommend something or get a better version.

Q : What about biases in algorithms? How do you address them?

A. In my case, I work with general algorithms or those applied to energy, where there are no obvious biases. But in other fields, such as algorithms that make decisions about people, biases are a problem. That's why there's a field called Fairness in Machine Learning , which seeks to reduce biases derived from unfair or poorly trained data.

Q. In what specific area are you currently applying the algorithms?

A. I'm in a lab working on communications and localization for mobile networks. I'm developing AI algorithms to locate devices, people, electric cars, or robots within spaces like factories, using multiple signals, such as Wi-Fi .

It's a field in constant development, but we're trying to improve it. I started my postdoc in February. We're working on it; I hope the paper doesn't take too long.

I'm concerned about increasing inequalities between countries that can access AI and those that can't. It's easier to manipulate information.

Verónica Álvarez, mathematician.

Q: And the data you use, where does it come from?

A. There are many public repositories with anonymized data, from different areas such as recommendations, movie reviews, etc. There's no way to know who's behind it. You can use that data to test algorithms. In general, privacy is highly protected. In Europe, they're passing legislation to regulate artificial intelligence, which I think is a good thing because it needs to be done.

Q. Do you think there is uncertainty about the future of artificial intelligence?

A: Yes. Many people are afraid of it, but I think it's a wonderful advance if used well. You have to think critically, of course. I'm worried about increasing inequalities between countries that can access AI and those that can't. It's easier to manipulate information.

Therefore, we must continue researching and ensure that development is ethical, but without hindering progress. We must unleash its full potential.

Q. In what other fields is it also being applied?

A. I think it can greatly improve our quality of life, from personalized medicine to everyday tasks. The emergence of ChatGPT is the latest development, and it makes writing an email or replying to a message easier for us every day.

Q. Now that you're in the US, what are the main differences with the type of research done in Spain?

A. Working conditions are quite improvable, due to low wages and very short contracts. After years of training, it's difficult to work with such instability. Many people leave because they don't feel valued.

I'll be here for a couple of years, initially. Then I'd like to return to Spain. I miss the food, the people, the everyday life.