895


Artificial Intelligence, a Colombian Perspective.

The need to integrate tools and machines to facilitate man’s work throughout the centuries has been a constant. Over 200 years ago, with the beginning of the industrial revolution at the end of the 19th century, automated machine systems were first designed and developed; they had a disruptive effect in the agrarian economic model of the time. In 1775 British inventor and entrepreneur Richard Arkwright, forefather of our modern factories, developed among other productive systems, an automated cotton carding machine model, this and many other innovations at the time contributed to the implementation of massive production systems and production lines in the textile industry. In time this generated among society, greater knowledge, technological progress, more demand for raw materials, better opportunities of mobility and labor supply, and in general terms social progress. Since then these ideas and systems have evolved and eventually became the industrial robots used nowadays in multiple industries.     

artificial intelligence

Source: Google images

 

R.U.RThe word “Robot” was used for the first time during Karel Capek’s play “Robots Universales Rossum (R.U.R.)” in 1921. The script introduces us to the character of R.U.R.’s general manager, Mr Busman who receives an unexpected visit from the president's daughter Helen Glory, to whom he explains “Robots are not people. Mechanically they are more perfect than we are, they have an enormously developed intelligence, but they have no soul.” To this she asks, “What do you make them for, then?”  Bussmann with a smirk on his face, answers her by asking the technical manager Mr Fabry, “What are robots for?” Looking at Helen, he answers “For work Mrs. Helen”. In a different scene of the play, another character explains, “Man shall be free and supreme; he shall have no other aim, no other labor, no other care than to perfect himself. He shall serve neither matter nor man. He will not be a machine and a device for production. He will be Lord of creation.”

 

While humans and the most sophisticated machines continue learning to be more intelligent, productive, improve our quality of life and mainly prolong our life expectancy, amid this new technological revolution, the collective ability to solve problems is ever increasing. A recent popular post in social media read that students should not be asked what they wanted to be when they grew up, but instead what problem they would like to help solve. This type of attitude when confronting challenges along with the massification of content, information, and knowledge can allow humans to be creative, innovative, and develop new ways to tackle problems that were previously inconceivable to solve. On the other hand, our capacity to learn and recall these massive amounts of information or big data is limited, and this is where technology can really be useful, for it gives us access to knowledge wich develops our own intelligence, and this is reflected in the behavior of the robotic machines and software being designed with Artificial Intelligence (AI) technology.

In the last 20 years, ever since IBM’s program “Deep Blue” was able to defeat Grandmaster Garri Kasparov in a series of official chess matches, Artificial Intelligence programs have evolved in such a way that the perspective of them surpassing human intelligence quotient (IQ) levels is already in the horizon.  Amazing software like IBM’s “Watson” which uses cognitive computing or Google DeepMind’s “Alpha Go” which uses deep learning  reinforcement learning, and recently beat the best “Go” player in the last decade, South Korean Lee Seedol.

Many disciplines contribute to the development of Artificial Intelligence. Since 1961 when the first patent for “Ultimate” was registered, a robotic arm capable of functioning in an automated way, technology has reached incredible milestones,  recently integrating technologies such as Natural Language Processing - NLP, Machine Learning, Big Data and thanks to the improvements in human - Machine interaction, every day it becomes easier to even tell the computers, in a verbal way, what to do.  In light of this, many questions and doubts arise, and there are plenty of dystopian stories in popular culture about circumstances in which machines dominate humans, even respected voices such as those from people like Elon Musk, Stephen Hawking and Bill  Gates warn us against the risks of a hypothetical scenario known as a technological singularity and Artificial General Intelligence.   

Source: YouTube

Surely these changes in society as a result of disruptive technologies will not occur all at once, but instead as a product of the numerous changes in the ongoing development of innovative technologies. So, in a slowly but surely manner digital immigrants and natives will continue to get used to the rapid and continuous changes happening every day, that way when a disruptive and meaningful idea comes along, people will not panic, but rather be more susceptible to accept changes in paradigms and other possible scenarios.

A type of automated program or software already doing its job on the internet is called, “Social Networking Bots” or simply "Social Bots”. This type of software uses Natural Language Processing - NLP and it’s designed to simulate human behaviors and interactions, with the objective of convincing people that they are interacting with a human, this in order to generate traffic in their websites or social network accounts, though there are also those that may breach our computer security with malicious intents, surely we must have read an article completely written by a software, and perhaps we may even follow a social media account administered by one of these tireless automated workers.

Massive media coverage during the most recent Consumer Electronics Show 2016 in Las Vegas, brought us the latest trends in robotic products focused on interacting with humans in their homes and workplaces, they are part of social robotics, we were able to see ideas such as a bartender machine, gesture-based controller gloves, a circular robot that can turn any wall into a video projecting surface, and even automated suitcases that can follow their owners around airports and bus terminals. One of the emerging technologies that showed its progress was facial recognition, and the multiple uses it can have in different fields. Searching about local initiatives in the field of Artificial Intelligence, we learned  that a couple of Colombian researchers received support from Google's Outreach program last year, this because of their research work carried out on tridimensional facial expression recognition.   

Colombian Accents had the opportunity to talk via Skype with Andres Romero, who studied bioengineering at the University of Antioquia and biomedical engineering, artificial intelligence, and computer vision at the University of Los Andes, and is one of the researchers benefited with Google’s stimulus program which fosters research in different areas of computer science around the world.   

With the intention of learning his perspective on these issues and other topics related to education, collaborative work, and the opportunities to access knowledge beyond the classroom, Andres Romero kindly shared with us his opinions in the following interview:

CA:     Hi Andres, thank you very much for being with us here in Colombian Accents. Andres, would you please tell us where you are from and where you grew up?

AR:     Hi, the pleasure is mine. I was born in San Marcos, Sucre. I lived in Medellin seven years. I have been living in Bogota for the past two years where I’m studying the biomedical and computer vision master’s degree program.

CA:     When you were in school, did you know what you wanted to study? Or, was there something or someone who influenced you to follow this path?

AR:     Yes, of course. Ever since I remember being very young, I always wanted to become a doctor, nothing else interested me since my father was a medical doctor, he’s a gynecologist, so that’s what I saw in my everyday life. I used to go and spend New Year’s Eve or a Christmas day at a hospital, I only wanted to become a doctor. As I grew up, I started to get familiar with mathematics and it became very easy for me to get into that world, and while looking into it, I found out that there was a degree called bioengineering, so I found out that in general terms it was a combination of medicine and engineering, so it rang a bell, also I read somewhere that it was the future, that one would build artificial hearts and lungs, and well, my parents always taught me to think big, not to stay behind or be restricted because of my place of origin, or for the lack of conditions that one may have, but instead, let’s think big, an artificial heart? What do we have to do? Well I chose this path of engineering and I think I made the right decision because I’ve been doing this for almost ten years.       

CA:     Andrés, in practical terms, easy to understand, please tell us about your work, what kind of technology are you developing, how long have you been working on this project? What have been some of the milestones reached so far, and how many people work with you on this program?

AR:     Professor Pablo Arbelaez and I are the only ones working, he is my tutor here for my master’s program’s thesis and it was also with him that we submitted or work to the Google program. My work in itself is to give an image to a computer and make the machine recognize the micro expressions associated with the face. The image obviously must be of a face, if the person is happy, well then it will certainly have the mouth slightly stretched towards the back and the cheeks will rise a bit, so that is the idea, to be able to detect the micro expressions associated with the face. In that way, there are many emotions and other moods that have more specific micro expressions; these exist in all faces, so it is very difficult for someone to hide these micro expressions because they are associated with all people. So that’s the idea, to detect these micro expressions and with them, to be able to determine if someone is sleepy, is telling the truth or if he or she is sad, and from then on basically whatever you can imagine. What we have accomplished so far in this project and it is extremely difficult to say, is that we are going to detect micro expressions and do it; it’s not as easy as it may sound. So, the first thing that we want to do is to detect important points on the face, for example we have three points on the eyebrow, on both ends and in the middle, we want to first detect those familiar points, and then see what happens around each one of those points. So far we can detect automatically 48 points on the face, on any face, weather it has an open mouth, it’s sideways or even turned.  What we want and is most important for the project is to see what happens around those points.

CA:     I imagine you must be a person who is always connected, always working with electronic devices, computers, tablets, with your smart phone. Do you frequently disconnect from it all? When you do, what other activities do you enjoy doing to relax a bit from your work?  

AR:     It is very difficult for me to disconnect myself, in fact right at this moment while talking to you, I have three screens in front of me here in my work station and I always keep an eye on it; it’s very difficult to disconnect. When I do, I really like traveling to places where there is none of that, no internet or cell phone signal, things like that.  That’s sort of my escape to put it one way, which is not often during the semester but I try to get away once in a while during a long weekend, get my backpack and go places where there is low connectivity, precisely for that reason.   

CA:     Andrés and with so much content available nowadays, how do you handle that? Because I imagine a lot of youngsters at schools today must have this sort of problem of being distracted by so much content that comes at us through the internet.  

AR:     Yes, I have the same distractions, I look at the same things, social media and all those things, but I think that one must try to handle these type of things. At one point or another during my day, I may be looking at social media, but I’m also aware that I have to be here and make progress on my project. I may be on social media but that doesn’t limit me either to be able to advance with this, the nice part about this area is that one is basically, in quotes, “teaching the computer how to learn,” so there are always periods of time when you may be waiting for a process to end, then you can take some time for recreation, to look for content, those sort of things. Yes, it is important to learn how to handle both things.     

CA:     What’s your opinion on the hypothesis of a possible technological singularity? What’s your opinion about a possible scenario like that, in our Colombian society?

            This is a topic that has been talked a lot about for a long time. The first application that was created in this area was around the year of 1989, when they wanted to automatize zip codes in the United States so that no person would have to manually classify the codes, and well that was the first time that it was done. Since then lots of changes have occurred, all in a slow and steady manner. So to the question that if one day, will the machines take control and destroy us? No, I don’t think so. Not in the near future, and I would say too far in a distance future, it would take a long time before anything like that may happen. To the question, will a machine take my job?, well that is another more reasonable question, and that’s what has already been happening. That’s a negative aspect. It’s a debate that must be held, if this may happen, what am I going to do? I believe it brings new questions, like, how creative am I? What other things can I do? Or, if the machine is going to replace me, am I going to be supervising the machine? That’s a legitimate concern.

CA:     What could be improved in Colombia to facilitate and foster research and development programs in areas as important as artificial intelligence and other branches of computer science?

AR:     I believe that motivation must be there since an early age, I think that a teacher from primary or secondary could radically make the difference, because most times one is taught not to try to change the way things are, or to become someone in life, but instead to try to go out and get a job. I think that, words of encouragement from a teacher could change someone’s life. That on the one hand, on the other hand I believe that here the way things are, it is very complicated to conduct research and budgets for research are very limited. This, lots of times instead of making one want stay in Colombia and work on research, it makes you want to leave. There are people like my tutor Pablo who precisely came back, he had been working at the University of California Berkeley and came back because he wants to contribute to research in this country and he believes and is convinced that there is talent here. This type of things also motivates me as a student and researcher to try to contribute to this country.

CA:     I imagine that in your line of work English language reading and speaking skills are very necessary, am I correct? And how are you coping with that aspect?

AR:     Nowadays it is completely essential to know English and much more in this area. The nice thing about this area, is that everything is open, if one wants to be taken seriously by the scientific community, one does the work and must release the code. This is my work and with this code it works, you try it  now, and everything is in English, all codes, when one goes to a conference, all forums, all tutorials that you find on the internet,  everything, absolutely everything goes in English. I have personally struggled a little because at the university where I studied my bachelor, English was not essential, it was a pretty simple exam at the end of the program, and it wasn’t that important. When I came to Los Andes University it was a total different story. Yes, I have struggled a bit because my tutor also has an American perspective and during our meetings everything is in English. Yes, it is something super important and I have had some difficulties with it, but I’m working on it.

CA:     Then, is it possible to find open source projects on computer vision for example? And are there support communities that can help someone access this type of knowledge?   

AR:     Yes, of course, there are several universities leading this area, among them the University of California Berkeley, Carnegie Mellon, Imperial College in London and all of them already have research groups on computer vision. They publish an article, the work they did, the code next to it and a forum. If it is not working for example, then you can write on the forum, this community is also very friendly, and they are always making sure that all goes well and moving forward, they respond quickly if something is not working well. That I believe is without a doubt the main reason why artificial intelligence has change so much in the last five years.   

CA:     Andres, thank you very much for been with us here in Colombian Accents. Finally, what type of skills would you recommend students nowadays to learn, to students from secondary for example, who may be interested in studying and working on projects as relevant and important as artificial intelligence?

AR:     Well, I’m going to paraphrase Pablo a little, the first time that I talked to him he told me that to work in this area one needs three things, first to know English, to like and know mathematics and third, programming. If you have those three, even at low levels, welcome on board. If you really think about it from the school you can start building these three. English at school, even if some schools are not bilingual like the one I attended, though I think my bases are already strong. Mathematics, it is very important to know mathematics in this area, if you don’t like it, it would be a good idea to do so, if you like this area. Programming, a lot of schools already teach programming even if it is simple code, there are lots of programming language codes, and tutorials on the Internet. You can find several free courses on coursera and edx.org many of those courses are free. You register and you follow the course weekly. Today knowing these three things is fundamental for this area.  

CA:     Andres, thank you very much for your time and for being with us on Colombian Accents.

AR:     No, thank you for the invitation and yes, I would like to inspire students who may read this, if you have a doubt, if you like what I do, please contact me without a doubt and I would gladly reply, and if possible I could be a guide on some points.

CA:     Is there a website, a blog, or an email address for someone who may be reading this interview and may want to contact you?

AR:     of course, our research group is called biomedicalcomputervision.uniandes.edu.co.

CA:     Once again, thank you very much for you time Andres.

AR:     Thank you very much!

Interesting links

 

  • Comments