Faces of Portuguese Research: João Gama

But long before AI became a cultural obsession, before language models entered our phones, offices, and dinner-table conversations, somewhere in Porto, João Gama was already helping to redesign the way machines learn from the world.

What makes João Gama’s trajectory particularly striking is not only the scale of his scientific influence, but also its timing. Many of the problems now central to artificial intelligence, adaptation, continuous learning, and real-time decision-making, were questions he had already been working on decades earlier. Quietly, from his lab at the University of Porto, he helped shape fields that only recently entered broader public awareness.

For João, machine learning was never something entirely new. “The term machine learning was first used in the 1950s,” he explains, referring to Arthur Samuel, the researcher who created a checkers program capable of improving as it played. “This ability to learn from experience is what we call learning.” What changed over time was not the existence of these ideas, but the scale at which society became aware of them.

A scientific ecosystem built in Porto

João Gama’s career became deeply tied to the University of Porto, where he began teaching in the Faculty of Economics in the early 1990s. At first glance, it might seem like an unusual place for one of Portugal’s leading artificial intelligence researchers to emerge. Yet João explains: “The group was founded in economics because that was where Professor Pavel Brazdil taught,” he says. Brazdil established one of the country’s first internationally relevant machine learning groups inside the Faculty of Economics itself.

In many ways, the location shaped the perspective, since economics and social sciences already depended heavily on large amounts of data and quantitative analysis. Traditional econometrics approached problems through predefined models; machine learning approached them differently, allowing systems to learn patterns directly from data itself. João found himself precisely at the intersection between computational rigour and real-world complexity.

Over the years, João helped consolidate that ecosystem through teaching, research supervision, and the creation of scientific networks that extended far beyond Portugal. He directed the Master’s programme in Data Analysis at the Faculty of Economics for more than a decade and supervised dozens of doctoral and master’s students, many of whom later built their own research careers.

Learning from moving worlds

João Gama’s observation that the real world is dynamic led to a significant shift in machine learning. While traditional models relied on static datasets, João worked with problems where data continuously evolved and became outdated quickly. This led him to develop approaches for learning from data streams, where algorithms adapt continuously in real time. At the center of this work was the idea of “concept drift,” the recognition that patterns change over time. From consumer behavior and urban traffic to industrial infrastructures and environmental systems, the patterns that shape the real world are constantly evolving.

João’s work emerged precisely from that instability, addressing a fundamental question that would later become central to modern artificial intelligence: how can machines continue adapting and learning within changing environments without repeatedly starting from scratch?

His work became foundational within the field. Today, his research on data stream mining and concept drift is among the most cited internationally, with tens of thousands of citations and a global influence that extends across both academia and industry.

The practical implications of this research became particularly visible through projects developed with real-world infrastructures. One of the most illustrative examples emerged through a collaboration with Metro do Porto, aimed at detecting mechanical failures before they happen.

Responsibility before superintelligence

Public conversations around artificial intelligence increasingly revolve around fears of machines surpassing humanity. João approaches these discussions with skepticism: “At this moment,” he says, almost humorously, “machines are still very stupid.” For João, contemporary AI systems remain fundamentally limited because they lack consciousness, self-awareness, and genuine understanding of what they are doing. They execute tasks with increasing sophistication, but without reflective awareness.

He does not dismiss the risks of AI. Instead, he focuses on present social realities such as inequality, data misuse, information manipulation, and unequal access to technology. He emphasises that technology is not neutral: AI shapes opportunities, labour structures, and information access, giving advantages to those who can work with it and leaving others behind.

This concern partly explains why he strongly values European efforts surrounding privacy regulation and responsible AI governance. João sees frameworks like GDPR not as bureaucratic obstacles, but as attempts to protect human autonomy in a world where information circulates at unprecedented speed and scale.

Universities as places of thought

Despite his international recognition, much of João Gama’s identity remains deeply connected to teaching and academic life. He speaks about supervising students with genuine affection, describing thesis mentorship as one of the most rewarding aspects of his career: “It is always good working with people,” he says.

This perspective also shapes his strong defence of universities as spaces that must preserve research alongside teaching. He believes universities are not merely institutions for transmitting existing knowledge, but are also responsible for creating new questions, cultivating critical thinking, and sustaining the intellectual freedom necessary for innovation. Research requires not only technical skill but also mental availability and the capacity to reflect on which problems are worth pursuing. Even after becoming Professor Emeritus at the University of Porto, João continues supervising students, leading research projects, and remaining actively involved at INESC TEC. In practice, very little changed besides stepping away from formal teaching. Research, mentorship, and scientific collaboration continue to occupy the centre of his daily life. What is particularly striking is the way he speaks about this trajectory with very little sense of individual protagonism. During his final lecture before retiring from teaching, João chose not to focus on distinctions, citations, or career milestones, but instead on the people who accompanied him throughout decades of work. “I did nothing alone,” he said. “I have to thank my teams, but above all, my students.”

Continuing to adapt

What becomes increasingly clear while speaking with João Gama is that adaptation itself sits at the centre of both his science and worldview. His research focused on systems capable of learning continuously because he understood early that static models struggle within dynamic realities. But the same idea seems to extend beyond algorithms. Throughout his career, João repeatedly positioned himself at transitions: between economics and computer science, theory and application, research and policy, academia and public debate.

There is something quietly consistent in the fact that one of the world’s leading experts in adaptive systems speaks so frequently about responsibility, collaboration, and collective learning. For João, intelligence, whether artificial or human, is never purely individual, but built collectively.

In the end, much of the infrastructure now underlying artificial intelligence, systems capable of adapting continuously to changing realities, exists in part because João Gama spent decades thinking about how to adapt to a world that never stops moving.