Sept. 18, 2023 — Artificial intelligence is changing the world at an unprecedented pace, driven by advances in supercomputing, the development of new algorithms, and the availability of vast amounts of data. This new paradigm highlights the need to reflect on the potential risks and challenges associated with AI, as well as on the possibilities and advantages of what has already been considered the next great revolution of humanity.
The Barcelona Supercomputing Center – Centro Nacional de Supercomputación (BSC-CNS) recently introduced a new book by Jordi Torres. The book was presented at the Centre de Cultura Contemporània de Barcelona (CCCB) to an audience of 400 people. Its goal is to explore the intriguing and promising journey of coexistence between humans and machines.
Artificial Intelligence Explained to Humans (Plataforma Editorial, 2023) is a work that explains the functioning and the economic, social and political impact of this technology in a precise and accessible way for readers without prior technical knowledge. The book is presented as a biography of AI that will allow us to form an opinion and have the necessary tools to participate in a debate on the limits and risks of AI, but also on its great potential to find new approaches to help us solve complex problems that until now have been difficult to tackle.
The presentation of the book at the CCCB was attended by the director of the BSC, Mateo Valero, the director of the BSC’s Life Sciences Department, Alfonso Valencia, the director of the Artificial Intelligence and Intelligent Data Science Research Centre (IDEAI-UPC), Karina Gibert, and the author, Jordi Torres. All of them debated the paths of this accelerated technological evolution and reflected on possible future scenarios in a conversation moderated by the journalist Pere Buhigas, which was part of the events prior to the exhibition ‘Artificial Intelligence’, produced jointly by the BSC and the CCCB, which will open on 18 October.
Supercomputing and the Role of BSC in the AI Revolution
The BSC director stressed the importance of supercomputing as the great driving force behind AI and the crucial role that the BSC has played in this revolution. “Chips, data and computers have brought AI out of its polar winter and this is changing the way we solve scientific problems. The current supercomputers we have at BSC have more hardware geared towards the development of AI systems, which will allow us to advance research in which supercomputing, together with AI, can be used, for example, to prevent and cure cancer in a personalised way, or to study and mitigate climate change,” Valero said.
The head of the BSC’s Life Sciences department has highlighted how increased computational power and massive data, combined with knowledge in biology, has made it possible to achieve scientific goals that until now seemed distant. “The potential of AI to solve complex problems depends on the quality of the data you train it on, and that is both its risk and its potential. AI, thanks to the data and the enormous computing power of today’s supercomputers, has allowed us to make progress in some fields in two years that we had not achieved in the last 30,” said Valencia.
The director of IDEAI-UPC insisted on the need to develop AI systems that are ethical, and therefore appealed to the ethics of professionals to control the misuse of this technology. “AI should not be autonomous, it should not make decisions on its own. It should solve complex problems, but the decisions are made by people, who are the ones who design the methods that we implement through algorithms. Unlike what happens in the US, in Europe we are clear that we must work to design AI systems that respect ethical principles and citizens’ rights,” Gibert said.
For his part, the book’s author pointed out that AI, like any tool with this potential, can be beneficial or detrimental depending on who is using it and for what purpose. “What is certain is that AI is proving to be an irreplaceable tool for accelerating discoveries and expanding reasoning and creative capabilities to address the great challenges we face as a society. The more astute AI becomes, the more it will allow us to use our brains in new fields, in new tasks, pursuing new limits and redefining the traditional concept of intelligence,” Torres stressed.
MareNostrum 5 and European Technological Sovereignty
One of the issues highlighted at the event was the need for Europe to create its own generative AI, those that generate images or text, such as ChatGPT, so as not to depend on the big tech companies in Silicon Valley. To do this, it is essential to increase Europe’s computational capacity with state-of-the-art supercomputers such as the MareNostrum 5, which is currently being installed at BSC.
The fifth version of the MareNostrum is a machine with different computational architectures, including a so-called accelerated partition that is specially designed to advance artificial intelligence and deep learning models. An example to understand the overwhelming evolution of these supercomputers in recent decades is that each of the 4,480 latest-generation chips that form part of the accelerated partition of MareNostrum 5, with a size of less than 10 cm2, will have twice the power of the entire MareNostrum 1 installed in 2005, which occupied the entire chapel of Torre Girona, some 160 m2, and was at that time the fourth fastest in the world.
MareNostrum 5 joins Lumi (Kajaani, Finland) and Leonardo (Bologna, Italy) as the only three European large-scale supercomputers. A number that BSC experts consider insufficient to advance Europe’s technological sovereignty in AI, given the high demand for its use by scientists and businesses. “If we want to have European sovereignty in AI development, we need to increase large-scale supercomputing capacity at the European level. We also need to design and manufacture European chips to create these supercomputers with our own technology and not depend, as is currently the case, on non-European companies,” said Valero.