Mário Figueiredo, our Senior Researcher, Distinguished Professor, and holder of the Feedzai Chair on Machine Learning at the Instituto Superior Técnico, recently spoke at the AI for Good Global Summit about one of the most important challenges facing modern artificial intelligence: understanding causality.
In his talk, Figueiredo emphasized that while current machine learning systems are highly effective at identifying correlations in data, they often fail to explain the underlying cause-and-effect relationships. Advancing methods that can uncover these causal mechanisms is essential for building more reliable and explainable AI systems.
The presentation focused on the emerging field of causal discovery, particularly on the challenge of determining the direction of causality from observational data - for example, distinguishing whether X causes Y or Y causes X. This problem becomes especially important when controlled experiments are impractical or unethical.
Figueiredo highlighted how progress in causal discovery could play a crucial role in the next generation of AI, supporting more transparent decision-making and accelerating scientific discovery across multiple disciplines.
The AI for Good webinar series “From Molecules to Models” is organised in partnership with the European Laboratory for Learning and Intelligent Systems (ELLIS), where Figueiredo is a Fellow. He also directs the ELLIS Unit Lisbon, a leading AI and machine learning network based at the Instituto Superior Técnico that brings together researchers from IT, INESC-ID, and ISR-Lisboa. For more information on the Unit, go here.
Listen/watch to the podcast: