Xavier Amatriain was born in Barcelona in 1973. His interest in technology and artificial intelligence (AI) and his ability to solve complex problems led him to become a renowned telecommunications engineer and AI expert. He has spent much of his career in Silicon Valley, USA, and is currently vice-president of Product AI Strategy at Google.

Previously, Amatrian had worked at Netflix, where he devised its recommendation algorithm system, and at LinkedIn, where he developed strategies linked to generative AI. He also co-founded Curai Health, which applies AI to healthcare. He has always been driven by a belief in the practical benefits of technology for society and has a solid grounding in research, having contributed to over one hundred scientific publications.

On 21 May he will return to the UPF, where he obtained his doctorate in Information and Communications Technologies in 2005. He will give a talk during the celebration of the 25th anniversary of Engineering studies at the UPF, which is the reason for this interview.

You will soon be attending the celebration of the 25th anniversary of the UPF School and Department of Engineering. You studied for your doctorate in Information and Communication Technologies there between 1999 and 2005. Since then, what do you think have been the main advances in these fields?

Technology is evolving exponentially. In other words, in the last 10 years it has evolved as much as in the previous 100 years. While this may seem like an exaggeration, consider that when I started my PhD, Google and Netflix, two of the companies that I've worked for and that are now part of our everyday lives, had just been created in the previous year and were completely unknown in Spain.

Information technologies in general and the internet in particular have transformed our society. And more recently we’ve entered the era of artificial intelligence, which will represent an even greater revolution.

You are currently leading Google's artificial intelligence and machine learning team. One of your main fields of interest is investigating the ability of artificial intelligence (AI) to solve complex problems. Could you give examples of complex problems that artificial intelligence tools can already solve today?

There are many examples of how AI can solve complex problems better than humans. A few years ago, at Curai Health, the start-up I co-founded, we demonstrated that AI can make medical diagnoses more reliably than human doctors. But today AI can go much further. Another example is AlphaFold, an AI application for scientific research, which Google Deepmind used to solve the scientific problem of predicting the structure of proteins in just four years. Until then, this problem had been considered practically impossible for scientists who were experts in the subject.

AI will lead to scientific discoveries that will radically accelerate medical advances in the coming years

How can artificial intelligence help improve access to health services for citizens?

AI can help the healthcare industry in many ways, but I would generally divide these improvements into three categories. First, it can help make better decisions and minimize medical errors: diagnostic errors cause almost 400,000 deaths per year in the USA. Second, AI can lead to much more efficiency, which will mean faster services and lower costs. Third, AI will lead to scientific discoveries that will radically accelerate medical advances in the coming years.

You also created the Netflix recommendation algorithm system. How do you think recommendation algorithms have evolved over the last few years to adapt their suggestions increasingly to the preferences of each user, and what challenges still need to be tackled in the future, including from the viewpoint of ethics or biases?

When I came to Netflix, there was no algorithm for products that used machine learning or AI. Today, advances in AI have directly contributed to recommendation algorithms that are much more accurate and powerful than those we had then. There are still many challenges and many of them are related to privacy and biases. To become more powerful, these algorithms need a lot of data. These data are extracted from user activity, with the risk that this entails, and in addition there are large collections of data that have many historical biases. But I don’t think these are the most important risks. The most important one is the ability of these algorithms to modify society’s behaviour.

Perhaps for the first time in history, users see AI as something they can “get their hands on”

At LinkedIn, you’ve also worked on developing strategies linked to generative artificial intelligence. This is the aspect of AI that has generated the most public debate since the emergence of tools such as ChatGPT. What are the main potentials and risks of generative AI today and in the near future?

This aspect has generated a lot of debate because it represents a revolution in the capabilities of AI, especially in the applications that users see. Perhaps for the first time in history, users see AI as something they can “get their hands on”. In addition, many have been surprised by the fact that this type of AI seems to have attained a certain amount of creativity and intellectual capacity. I mean, it’s very different from the AI we expected, which would contribute to self-driving cars and robots that cook for us. Like any technology with great potential, it comes with great risks. I think the biggest one is that we leave its development in the hands of actors who have negative goals and values, such as terrorist groups and countries with dictatorial or extremist governments.

Companies and organizations that don’t adapt to AI will be so inefficient that they’ll be replaced by new ones

How do you think companies and organizations can harness the potential of AI to improve the efficiency of their practices and procedures or to promote innovation?

Companies and organizations today must rethink absolutely all processes and practices to adapt them to the new reality of AI. In the same way that no one would think of working without the internet now, it will soon be unthinkable to work without artificial intelligence. Companies and organizations that don’t adapt to AI will be so inefficient that they’ll be replaced by new ones. For example, computer programmers who use AI will be at least one hundred times more efficient than those who don't.

From your experience as a teacher, do you think universities will have to adapt their teaching and research methods to the age of artificial intelligence?

Absolutely. In fact, I think most are already doing so or at least considering it. The first reaction, as could be expected, was to try to ban its use. It reminded me of the time when I was at university and you weren’t allowed to use Wikipedia as a source of information. But I think it’s now quite clear to everyone that this revolutionary technology is here to stay and must be taken advantage of. I’ve already had conversations with university professors in the US who not only allow but require their students to use AI and expect the assignment to include a detailed report of how they used it.

Finally, what initially led you to dedicate yourself to the field of engineering and what continues to drive you to work on cutting-edge projects in this field?

I’ve always liked to learn. Working on cutting-edge projects and new technologies allows me to earn a living while continuing to learn every day. Also, because I’m an optimistic person, I believe in the value of science and technology to improve the world. No one can deny that today we live much better than in the past. I think this is largely due to scientific and technological advances, and I like to think that I’ve contributed to it personally, even if only minimally.

We need (...) a greater diversity of people to help us design technologies for everyone and avoid biases resulting from the lack of diversity of technologists

Why and how would you encourage young people, and particularly women who’ve traditionally been a minority in this field, to study engineering and pursue a career in this sector?

It’s very important for us to encourage women and minorities to pursue technological studies and fields. Technology is the future, and we need not only more people working in it but a greater diversity of people to help us design technologies for everyone and avoid biases resulting from the lack of diversity of technologists. In fact, a few Catalan technology leaders have launched an NGO called Friquifund with the idea of facilitating technology education for at-risk minorities. We need to make technology a tool that reduces inequalities and increases opportunities for everyone, but this requires work and action in this direction.