Nuria Oliver, who lives in Redmond with her husband and a two-year-old son, says that being Spanish and young has not prevented her from reaching high in an environment as competitive as science in the United States. The proof is that in 2004 MIT included it in the list of the 100 young people who are in the world forefront of research on computers, nanotechnology and biotechnology. His favorite field of work is intelligent machines that will one day make the leap towards emotions.
–There are two main pathways to artificial intelligence: mimicking the brain with neural networks and creating learning systems. Which is better?
-Neural networks are based on the idea that if we understand the mechanisms of the human brain we can reproduce them in a machine, but the brain is very complex and copying it is very complicated. For me, the second way is better because it gives more practical results. On the subject of speech recognition, for example, we can collect a lot of data and look for algorithms that solve the problem without the need to reproduce human mechanisms. These learning techniques that do not reproduce biological models are called Statistical Machines Learning.
–Are they separate paths or do they converge?
-I think it’s good that they contrast. The normal thing is that if a learning system works, you then go to the biological model and compare it. There are many things that we do not know, for example, if the brain also uses, internally, systems of this type that facilitate its work.
–Many artificial intelligence systems work, but in isolation: speech recognition, handwriting, vision systems … How difficult is it to integrate them?
-Yes. Artificial intelligence began by promising too much and that is a problem that continues to drag. These days you can talk to computer systems and they are very effective about what they know, for example booking airline tickets, but you can’t talk to them about anything else. They work on very specific domains. We are very far from being able to pass the Turing Test – a machine that in normal conversation manages to pass for human – and from an artificial intelligence with something similar to what we call common sense.
–Must machines have sensitivity to be smart?
-They must have the ability to perceive and transmit emotions, which are a very important component of our intelligence. To begin, they will have to understand, through sound and image sensors, what happens to us, understand if we are sad, happy or confused and communicate with us based on that information.
–So the machines will adapt to us, instead of us to them …
-That’s how it is. Machines always behave in the same way, they are totally predictable. They do not understand what we want and they are hostile to us because they insist on an answer that does not solve the problem at hand. Through perceptual intelligence, it will be the machines that understand what we ask them to do.
–Are movies about artificial intelligence, like AI or Me, Robot, far from being reality?
-The movies are movies, but they visualize some aspirations. In the case of humanoid robots, of which there are already some interesting models in Japan, the best thing is that they have bodies, so that they can interact with the environment as a human does and learn to move, not to stumble … Ultimately it is a way for the artificial and biological models to work together.
–Talk about learning, but any artificial intelligence system must seem foolproof …
-One of the problems with these systems is that they have to be robust and work with all users and in all situations. If a machine tries to be smart and gets it wrong, it creates great confusion. While working with a dumb machine you don’t expect great things, but a mistake in a smart machine makes you distrust.
–Are you also researching mobile phones to be something more than that?
-Yes, I have chosen mobile phones because they have more and more computing capacity and because in many areas, such as South America or Asia, they are the most popular computer. My idea is to make the mobile, through artificial intelligence, understand where the user is – in the car, at work, at home – and detect their mood to better communicate with them. This can be achieved with cameras, microphones, speakers -which are already in many mobiles-, motion sensors … Another project is the monitoring of our biological signals. Physiological sensors capture vital signs, and artificial intelligence interprets the signals and finds correlations between changes and factors such as stress, physical exercise, diet or hours of sleep.
–A telephone that will communicate us with the interior instead of with the exterior …
-We have a lot of information in general, but very little about ourselves. Many people go to the doctor only when they are sick and there they take their pulse or pressure, but there is no data on how they have evolved in recent weeks. It is not about replacing the doctor, but about providing more information.
–You have expressed your concern about the little support that the investigation has in Spain. How do you see the situation now?
-The socialist government says it wants to promote research and there are projects in this regard. I know that they want to re-fish researchers who are outside and it seems that they are aware of reality and are willing to invest to change the situation. I have hopes.
–Personal or collective?
-Both. Changing the situation in Spain is complicated because there is no research culture and that does not appear overnight. You have to create the mentality, the infrastructure and constant support for the investigations. What happens when the money runs out, or when the government changes? What I would like the most is to do in Spain what I do here.
–Is the unity of Europe positive for Spanish science?
-I hope so. In Europe there are important centers such as the Max Planck Institute (Germany) or the Inria (France). Europe, and within it Spain, must research and develop its own technologies. The problem is that there is a lot of difference between the American style and the European style. In the United States the attitude is more practical; you try to do something, period. In Europe, before starting a project, all its philosophical consequences are considered. But advances in technology go very fast and if you spend a lot of time thinking about whether to do it or not, when you decide it is already done or the idea is obsolete.
–Without failure there is no science?
-That’s how it is. In the US that is widely assumed. There are many small companies that risk their capital, fail and it is not the end of the world, because then their managers promote another project. That’s another big cultural difference. In Spain there is a lot of fear of failure, and you cannot progress in research without making mistakes.
–You are a woman, young and Spanish. Has it been a disadvantage?
-I have never felt marginalized or discriminated against, but it makes you strange to see that you are the only woman. My area of research is dominated by men and the number of women studying technical careers in the US has dropped alarmingly. I recently gave a talk at the University of Washington and there was not a single female student in the audience.
–How do you explain this setback?
-It’s something social. Models presenting TV series, movies, and education are almost never female researchers. And the student who starts, being often the only one in the class, is uncomfortable and ends up leaving him.
–The president of Harvard University recently said that women were not appropriate for technical careers.
-No comment. Anyway, it is a complex issue. Engineering or computer science majors have a reputation for being unsocial and don’t attract many women. It seems that there is a cultural correlation between career and gender, and that women prefer medicine or others in which there is more human interaction. But the shortage of female students in technical careers is negative for the scientific community. It is not good for one sex to dominate.
Alex Barnet
