Artificial intelligence, word of the year 2022 by the FundéuRAE

The FudéuRAE, Fundación del Español Urgente chose this expression because of its media impact and the debate around its use in multiple areas

At Dartmouth College (New Hampshire, United States), John McCarthy, Marcin Minsky, Nathaniel Rochester and Claude Shannon started talking about artificial intelligence (AI) in the summer of 1956. During all these years, artificial intelligence has gone through times when its relevance and visibility have been in the background. The research was relegated in previous times because it failed to answer most of the challenges that initially were presented.  

We interviewed Prof. Jose A. Lozano, Scientific Director of the Basque Center for Applied Mathematics - BCAM and Professor of Computer Science and Artificial Intelligence at the University of the Basque Country (UPV/EHU). Professor Lozano has built his research career on delving into different aspects of artificial intelligence.

The development of artificial intelligence in recent years rests on the exponential increase in the computational capacity of computers. It has made it possible to create models that perform activities more accurately and faster than humans, such as classifying images. "AI lives with us in many areas of our lives. Starting with mobile apps, Spotify itself is an AI-powered song recommendation system. Along these lines, the voice assistant Siri in the iOS operating system is also an artificial intelligence system, as is chatGPT, a chat system based on the GPT-3 AI language model. In the automotive field, autonomous cars are based on using artificial intelligence tools and models. In some cases, the design of bus routes is also based on artificial intelligence techniques. In leisure and recreation, online chess is an example of how digital players have developed its own play, without having seen any human moves, for example, AlphaGo and AlphaZero," says Professor Lozano.

At BCAM, we are working on several research projects in Artificial Intelligence. On health, we are working on cost prediction, treatment prediction and predicting the severity of a patient with SARS-COV19. In the field of energy, we are researching demand prediction. In this last project, the results obtained have recently been published in an article, receiving the Best Applied Contribution in Statistics in the SEIO-Fundación BBVA 2022 Awards.

Another area of artificial intelligence that is researching at BCAM is Fair Learning. "Most of the biases that artificial intelligence systems have are because the data set used to learn the model is already biased from the begining. The aim is to learning systems that do not discriminate based on sensitive attributes such as gender, race, or religion. To this end, we have created the BCAM Severo Ochoa Strategic Lab on Trustworthy Machine Learning, collaborating with Prof. Novi Quadrianto of the University of Sussex (UK).

"Adversarial learning tries to create instances that go unnoticed by the user, but confuse the system. These instances represent a breach in the security of artificial intelligence systems. For example, in autonomous cars, it is possible modify a single pixel of a traffic sign and have the automatic recognition system confuse it with another one. In this research, we create attack methods and defences for the system,' says Professor Jose A. Lozano.

One of the peculiarities of BCAM is that we do research on the creation of the mathematical basis of artificial intelligence. At the same time, we emphasise its practical application in the field of industry and security working together with the BAIC, the Basque Artificial Intelligence Center, and the BCSC, the Basque CyberSecurity Centre. "Our research is applied, for example, to detecting anomalies in machines or predicting the lifetime of devices.

As far as the future is concerned, Jose A. Lozano explains that the possibilities are vast. "Right now, we are surrounded by systems with AI. However, this is only the beginning. Artificial intelligence can revolutionise the entire field of healthcare, making personalised medicine a reality." Artificial intelligence could have an application creating systems to help humans, for example, the elderly people. "This group needs to be accompanied in various activities, and although there is no substitute for human companionship, automatic systems can help and above all prevent.