As Head of Responsible Artificial Intelligence at Meta, he has led initiatives related to the robustness, transparency and security of AI, which are key to the development of the technology company.
At a time when supercomputing centres are becoming AI development centres, Canton's appointment will help place the BSC at the forefront of AI research globally.
Researcher Cristian Canton will be the new Associated Director of the Barcelona Supercomputing Center – Centro Nacional de Supercomputación (BSC-CNS). Canton (Terrassa, 1980) has previously served as director of Responsible Artificial Intelligence at Meta's headquarters in the United States, and will take over from Josep M. Martorell, who is leaving the centre after a nine-year career during which BSC has established itself as one of the most prestigious supercomputing centres.
PhD in Computer Vision from the Polytechnic University of Catalonia (UPC) and expert in machine learning and artificial intelligence system security, Canton thus closes a period of more than 15 years in large corporations, during which he has participated in multiple initiatives related to the robustness, fairness, transparency and legitimacy of AI.
At the company led by Mark Zuckerberg, Canton has played a key role in defining the strategy for developing secure and responsible AI systems on platforms with a broad global impact such as Facebook, Instagram and WhatsApp.
He began his career in 2016 in Facebook's Community Integrity division, focusing on harmful content, misinformation and manipulated media, with a special focus on issues such as child safety and terrorism. In 2020, he created the industry's first AI Red Team to test the robustness of AI systems and understand their vulnerabilities and risks related to misinformation and election interference, among other issues.
Since 2021, Canton has led Meta's Responsible Artificial Intelligence area, coordinating the company's efforts in equity, robustness and security, governance and accountability, and transparency, especially in the field of Generative AI. During this time, he played a key role in the development of successive versions of the Llama language model, one of the most advanced and efficient freely available models for research and non-commercial purposes, to ensure its reliability against potential misuse, bias and security threats.
Before joining Facebook, the new associate director of BSC worked as a researcher at Microsoft Research in the US and UK, and at the motion capture systems company Vicon, also in the UK, where he contributed his knowledge of computer vision to produce visual effects for the film industry.
Promoting research in AI applied to multiple disciplines
At a time when supercomputing centres are becoming AI development centres, Canton's appointment will give the BSC a boost to position itself at the forefront of AI research applied to multiple disciplines. Currently, BSC has more than 320 researchers working on the development of AI projects aligned with European values in fields as diverse as biomedicine, finance, climate and public services.
Canton will succeed Josep M. Martorell, who has served as Associate Director of BSC since 2016. Martorell, together with Mateo Valero, Director of BSC, has led almost a decade of growth in which the centre has opened a new corporate headquarters, installed the fourth and fifth versions of the MareNostrum supercomputer, tripled its workforce to over 1,300 employees, and created 14 spin-offs in its commitment to transforming scientific knowledge into real impact for society.
During this time, BSC has managed to ensure the sustainability of major projects, such as the BSC AI Factory, which aims to drive European innovation in AI in strategic sectors, and the MareNostrum 5 supercomputer itself, which will soon increase its processing capacity to respond to global AI challenges.
This article was first published on 2 June by Barcelona Supercomputing Center.