“Humans need to learn how to interact with smart machines”

Simon VuillaumeDirector for International Projects, Cegos Group

Manuela Veloso, professor and Artificial Intelligence specialistRead the interview of Manuela Veloso, a woman of science who was part of the first machine learning department in the world. She is the co-founder of the International RoboCup Federation and researcher in the artificial intelligence field.

Currently a professor at Carnegie Mellon University in the U.S. and one of the first women to work in computer science, Manuela Veloso speaks about the future of artificial intelligence and the gender gap in the world of the STEM (Science, Technology, Engineering and Mathematics), among other topics.

>View the original article published on August 9th, 2017 by Link to Leaders (in portuguese)

You trained in electrotechnical engineering at Instituto Superior Técnico de Lisboa and after completing your doctorate in computer science you joined Carnegie Mellon University in Pittsburgh, USA in 1992. A Portuguese woman in the international world of technology when technology did not yet have its current dimension. What were the biggest challenges that you faced at the time and what did you learn?

I have been living in the United States for 34 years. Ultimately, it involved starting a new field of study, seeing as computer science was significantly different from electrotechnical engineering and in the USA it involved registering for courses, doing research. It was fascinating doing research for my doctorate thesis at Carnegie Mellon because in the end it was a whole new field for me and I was at a place with enormous amount of dedication to state-of-the-art research. It was a much more rigorous system in terms of courses and credits, seeing as there was a lot of homework with very tight deadlines and I was not used to that, but I eventually got used to it.

The field you work in is male-dominated, right?

Yes, all of science, including the so called STEM, namely science, technology, engineering, computer science, all of those areas are male-dominated.

What else is needed for better gender equality in this field?

Essentially, what is needed is for us to say that there are not any predetermined professions. Women, girls and children must believe that everything is within their reach. I grew up liking mathematics and whether it was at home growing up or in my in high school environment, no one said, “Mathematics and engineering are not for girls.” No one ever said that to me. Basically, stereotypes have to be eliminated. I went to Dona Leonor High School, which at the time was an all-girls school and once in a while our graduating class gets together and we are a group of 60-year-old women, a great group and we all get along well. Not many of us are engineers, but there are physicians, economists, teachers and a few engineers – it is really impressive.

Was it a successful generation?

Yes, it was. We are all women who are extremely dedicated to science. There are also some who chose literature, but all in all, it is a fantastic group of women. We all grew up in an environment where our families never said we could not do anything because of gender. Here in the USA, today there is an attempt to increase the number of women in technical fields. At Carnegie Mellon our Computer Science course has continued to increase the percentage of women who graduate in the field. The next academic class beginning in August will be made up of 49% or even 50% of girls. This is really great. It is the only place in the world where the percentage of girls admitted to the computer science program is 50%. It really is great and here at CMU (Carnegie Mellon University) there has been a significant effort made toward this achievement. We have an association called Women at SCS (School of Computer Science) led by a female professor. In the doctorate and master’s programs the gender distribution is not 50%, but the percentage is significantly higher than at other institutions.

Artificial intelligence is currently one of the hot topics. What do you foresee for this field in the near future?

I would like to reiterate an idea that I will also touch on at the Business Transformation Summit in October. This is the situation: artificial intelligence (AI) is an area where I have always done research and AI essentially consists of the capacity to have computers analyze information and make decisions, to have them learn. Computers that are capable of being intelligent. I am talking about a very broad spectrum of rationalization. This area is crucial because our society increasingly stores data. Everything is digitized, we take photos, make videos, everything we are currently experiencing, our GPS, our mobile phones, everything is digitized and becomes information stored in a computer. As such, with all this information that we are accumulating, it is impossible to think that we will look at all the pictures and videos, at all of it and be able to process and derive interesting conclusions. Therefore, AI is the vehicle that will determine that this information will be processed by algorithms. I think AI will be fundamental if it is linked to data, i.e., to information, to everything that is linked to computers.

You mean there will be some development with regard to data processing and information?

Exactly, a very major development, given that this data is currently being processed, but can also be processed at a completely different level, used to support human decisions and make previsions. It will be a whole other type of processing when we have more AI to interpret, understand and use the data that we collect.

What risks should be guarded against with this rising interest and development in AI?

Essentially, everyone says, “It will make people lose their jobs,” but I think that once we have begun going down this road we cannot go back. When computers began to be available to the whole world and everyone began to program and develop all types of algorithms, it was something extraordinary. We cannot think in any other way, other than that we have to educate people. People have the responsibility to use AI for society’s benefit.

We know that any technology has good and bad uses. IA is a technology that was invented by humans, so it did not fall from the sky from who knows where. It was invented by human brains and therefore people have a responsibility to use it properly. AI is known as the Fourth Industrial Revolution, but as with any industrial revolution, the Fourth Industrial Revolution is humanity’s responsibility. The challenge is to have mechanisms that educate people on its proper usage and to ensure that people make proper use of the technology.

You developed the CoBot service robots, autonomous and intelligent, capable of detecting their own weaknesses and asking for help. What are the greatest challenges you faced when developing these robots?

This concept of asking for help was in fact very new. The interesting part is having robots move around inside the buildings. There is the technical part of making sure they move safely, i.e., the more technical part of the motors, sensors and planning algorithms. In other words, there is this challenge, but there is also a learning process in order to increase their capacities over time by interacting with humans and asking questions.

You are also a co-founder of the RoboCup Federation. Why was it started?

RoboCup is quite fascinating because first of all, it allows us to research autonomous robots. The fact that there is an adversarial team with a football problem and we have to develop algorithms in a situation where we do not know what the other team will do from a strategic point of view, or in terms of perception or action, it is not always easy because it is difficult to plan when surrounded by uncertainty. The first thing involved this problem of the uncertainty surrounding the adversary, while another thing involved the fact that robot football has allowed research to be conducted for the first time in a very concrete manner on autonomous robots that work together as a team. Football robots have to attain objectives, such as scoring goals as a team, and this is where the multi-robot part comes in, that is to say, having algorithms in which the tasks are distributed among robots.

Seeing as you are one of the speakers of the Business Transformation Summit in Portugal, what are the main messages that you will be conveying?

I will attempt to explain some more about what AI is because even though a lot of people talk about AI, I am convinced that people really do not know what it means from a technical point of view. Therefore, I am going to first explain what AI is in the world today and in terms of existing data. Then, I will also be talking about Human-AI interaction, which is the interaction between humans and artificial intelligence. I will be presenting my theory that this interaction between humans and AI is an interaction of cooperation, and humans have to understand how it should be used and how to interact with these smart machines. However, there is also a need for smart machines to be much more transparent for humans. I will end my presentation by explaining how I have developed methods for making my smart robots more transparent, explaining to humans the decisions that the robots make, while also answering questions that humans may have regarding their autonomy.

Which sectors are more receptive to AI?

I think the financial sector is very receptive, along with the retail and health sectors. Processing health data, heart data, brain data, i.e., health treatment is a very receptive area. Then you have energy and climate, namely the areas in which people are more directly included.

Business Transformation Summit

Manuela Veloso presented a keynote at the 2017 Business Transformation Summit

Read our post on "the future of machines and human" to discover the key takeaways.

Written by

Simon Vuillaume

Simon used to share his insights in this space when he worked at Cegos Group
Learn more
newsletter image

Receive our newsletter

Keep up to date with the latest L&D Insights

Subscribe here