When artificial intelligence overtakes us

Beyond the immediate advantages of artificial intelligence, today its accelerated development poses serious ethical and legal challenges. In this report, the Information Agency Conacyt documents the most outstanding results achieved by Mexican scientists in the field of artificial intelligence, in addition to investigating its implications in our daily lives and work.

The development of computers equipped with programs capable of emulating qualities that define human beings -such as intelligence, creativity, and learning- is no longer a subject of science fiction, but a present reality that is manifested in thousands of algorithms that evaluate our searches, tastes, and habits.

In 1993, Rafael Pérez y Pérez became a father for the first time, his "baby" was born and a few months later he abandoned him because he did not meet his expectations, he wanted a gifted child, who was capable of doing things that had never been seen before. A year later, he started planning another child, this time he would be more careful in the whole process, make better choices, and make better decisions. The planning process took him about three years.

After four years of being a father for the first time, Rafael Pérez had his second child. He was really excited, he expected this one to be very intelligent, so from the first moment he had contact with him he took great care of him, every night he reads him something, sometimes stories, sometimes literature or poetry. He has the illusion that as he matures he will become a writer recognized for the quality of his stories since he always wanted to dedicate himself to music, but life took him down the path of computers.

To ensure that the little one will have a successful future, Rafael Pérez y Pérez, a research professor at the Universidad Autónoma Metropolitana (UAM), subjects him to constant hard training in which he uses a learning method called engagement and reflection. It seems that this training and discipline of reviewing everything he learns, through stories that Rafael Pérez tells him every day so that the little one can generate ideas and then create stories, which are evaluated by the child, keep the good stories and discard those that are not and learn from those that are creative and novel to create more, has given him good results.

Currently, Rafael Pérez's creation is capable of writing more than 400 stories in less than six seconds. While not all stories are outstanding, at least 70 percent are worth keeping for evaluation. As the years go by, his stories are better, more polished and his language is greater; also, his abilities to make stories more coherent and creative are growing.

The little Mexican is causing a stir in various parts of the world. In fact, several experts in artificial creativity in the United States and some European countries are paying attention to his literary career. Rafael Pérez y Pérez is very proud of the little boy and wants to celebrate his son's first 20 years with the publication of the book Mexica: 20 años, 20 historias, which presents 20 of the thousands of stories the young man has created.

This is not just any book, people from all over the world were hoping to read the stories (written in Spanish and English) of this Mexican promise, even the book has a preface written by the prestigious scientist of the Massachusetts Institute of Technology (MIT), Fox Harrell. This is one of the 20 stories that the young writer presented in that text:

The knight ocelot was the princess' father.

As part of a plot, the enemy kidnapped the princess and planned to kill her at nightfall.

The knight ocelot knew that the gods expected him to free the princess.

Suddenly, the knight ocelot met the enemy. He couldn't believe it!

Annoyed, the princess humiliated the enemy.

The knight ocelot looked at the enemy and attacked him.

Determined, the enemy took a large stone and dislodged the knight ocelot.

The princess administered the ocelot knight the potion she had prepared. He got better quickly.

Suddenly, the knight ocelot and the enemy engaged in a hard fight.

Without hesitation, the knight ocelot took every last breath from the enemy.

The knight ocelot cut the princess' ties. Free at last!

This story is one of the thousands written by the computer system that generates narratives, called Mexica, which uses artificial intelligence (AI) to create stories, an activity that until a few years ago was thought to be exclusive to human beings. This activity raises questions and challenges: are machines really creative, will machines outperform humans, will they take away our work, will humans be enslaved by machines, will machines act according to ethical rules?

The machines, creators, or imitators?

All the above questions generate a lot of controversy and heated debates among experts, while in the rest of the population they generate uncertainty and fear, the latter feeling often originated by films. On the one hand, a sector of specialists in artificial intelligence states that machines are not intelligent and that they cannot yet create things, since they do not have the creative capacity and only follow programmed instructions, therefore they are simple imitators of masterpieces made by humanity.

"It is easy to have ready-made sentences, combinations, and rules to make the combinations and then take out a poem, but the computer has no idea what a human can feel when he or she comes into contact with artistic work, for that you need empathy and to assume what that person is going to feel, that cannot be done by the computer," says Raúl Rojas, a research professor at the Free University of Berlin.

The problem, he says, is that there is a lot of exaggeration about artificial intelligence and both the press and marketing always present something that apparently machines can already do. "When those of us who work in this field see what exactly is happening, we know that it is not true, it still does not compare to what a human does, especially in activities that have to do with intuition, feelings or emotions; there is no computer yet that can do this kind of thing".

The game of imitation

In the same vein, Humberto Sossa, a prestigious research professor at the Computer Research Center of the National Polytechnic Institute (IPN), indicates that so far no software can make a masterful artistic work, but progress is being made in this direction and perhaps in the future, it will be able to reach human talent.

"Artificial intelligence can make some works, but there is still a long way to go before it can make work like Mozart's, up to now they are imitators, not creators, they are two different things, for a computer to have its own creativity there is still a long way to go (...) It is true that there are already things that are going in that direction, but we could not yet speak in the short term that a machine is going to achieve it".

While some specialists point out that artificial intelligence cannot yet create anything original, there are others who say that computers are a very useful tool for better understanding how the creative process works, both in humans and in machines. "Computers, like human beings, can generate novel products by learning, imitating, making mistakes, evaluating and improving; nobody creates something masterful at first," says Rafael Pérez y Pérez.

This is just the beginning, and perhaps Mexica's main contribution, since it gives details or approximations of how the creative process may work in humans, stresses Rafael Perez, who is also president of the Association for Computational Creativity. In fact, this Mexican innovation, which was explained in the specialized article MEXICA: a computer model of a cognitive account of creative writing, published in the Journal of Experimental & Theoretical Artificial Intelligence, in 2001, adds to date 176 citations, an outstanding figure in the area of computational creativity.

As well as Mexica, there are other computer systems that are creating pieces considered artistic, for example, DeepBach, developed in the Sony Computer Science Laboratories in Paris, this has managed to compose melodies so harmonious that 50 percent of the people who heard them -some professional musicians- thought that they were creations of the famous German baroque musician Johann Sebastian Bach, as reported in the scientific article DeepBach: a steerable model for Bach chorales, published in 2017.

The age of intelligent machines

Although the scientific and technological advances cited are encouraging, experts such as Humberto Sossa and Raúl Rojas consider that artificial intelligence is currently in the so-called weak or applied phase, since it resolves specific tasks, focused on helping human beings, but does not attempt to simulate the full range of cognitive abilities of living beings. This type of artificial intelligence is present in everyday life as a tool that facilitates tasks for people and improves their quality of life. For example, companies such as Google, Amazon, Netflix, Facebook, Spotify, and Waze, among others, use artificial intelligence to offer us better services and satisfy our needs.

What this intelligence does is "follow the thousands of tracks" that we leave when we surf the Internet every day so that with all the information (big data) they collect, they can analyze it and learn from you, what you like, what interests you, where you live, where you work, and they might even know if you're pregnant. Artificial intelligence knows you so well that at any moment it can give away your darkest secret, as revealed in the text "How companies learn your habits", published in The New York Times, in which it narrates that a father knew his daughter was pregnant because of the personalized advertising that arrived in his mail, based on the searches, browsing sites and purchases of the teenager.

With all this, we should already be worried and busy, since artificial intelligence knows us so well that it can persuade us to consume a certain product or service, and it could also influence our preference for a particular candidate during an electoral process, as in the world-famous case of Cambridge Analytica. But this is just the beginning since the examples mentioned are in the world of the Internet, but in the next few years, artificial intelligence will be circulating on the streets aboard autonomous vehicles.

The machines could take your job away

In addition to autonomous vehicles, in the next few years, artificial intelligence will have a greater presence in almost all jobs and will even make some of them obsolete, Humberto Sossa points out. In the last 10 years, machines have been trained to solve increasingly complex problems. Today, machines are capable of carrying out tasks once only attributable to human beings, such as diagnosing an illness, writing a newspaper article, determining whether or not someone will be given credit, or identifying people in the middle of a crowd, among many others.

According to estimates made by the international consulting firm Gartner, by 2020 artificial intelligence will have eliminated 1.8 million jobs worldwide, one of the industries most affected being manufacturing. Likewise, this study conducted by this consulting firm points out that in 2022, one out of every five workers will perform non-routine tasks and will have the help of artificial intelligence to do their job.

How will artificial intelligence impact Mexico?

Mexico is not alien to this environment in which automation and artificial intelligence will impact jobs, in fact, in the country the forecasts of unemployment generated by intelligent machines are more drastic since the industry most affected is manufacturing and Mexico is a manufacturing country.

According to the study Towards an AI strategy in Mexico: harnessing the AI revolution by the Mexican federal government, Oxford Insights, and C Minds, artificial intelligence will impact 19 percent of jobs in Mexico. "9.8 million jobs will be affected by automation over the next two decades, ranging from tasks that are facilitated by automated systems to jobs that will be completely replaced," the study said.

Is Mexico prepared to deal with artificial intelligence in the coming years?

It is impossible to know what will happen in Mexico in the next few years and even more so when the country faces a political transition. Notwithstanding the scenario, the issue has already been placed within the National Digital Strategy. In fact, the study Towards an AI strategy in Mexico: harnessing the AI revolution, in which the government, academia, civil society, and the business sector all participated, points out where Mexico is, what its strengths are, and areas of opportunity.

Yolanda Martinez Mancilla, the coordinator of the National Digital Strategy, explains that derived from this study, which was supported by the UK Embassy in Mexico, the AI plan should focus on five guidelines: government and public services; data and digital infrastructure; research and development; capacity, skills and education; and legislation and ethics. Perhaps if in the coming months a national plan for artificial intelligence was established to know what to bet on as a country, Mexico could play an important role in this technological area.

According to the Index of AI preparedness for government, conducted by Oxford Insights, Mexico ranks 22nd among the 35 countries of the Organization for Economic Cooperation and Development (OECD). This means that Mexico scores well in terms of digital infrastructure and open data policy. But it has areas of opportunity in terms of technological skills.

People, Mexico's main capital

For Mexico to develop its scientific and technological skills, it needs highly trained people, especially if it is already known that many manufacturing jobs will be lost. Miguel González Mendoza, a research professor at the Monterrey Institute of Technology and Higher Education (ITESM) and president of the Mexican Society of Artificial Intelligence, stresses that Mexico must bet on the formation of talent.

"In reality what we need is the construction of capacities in the generation of talent. It is necessary to train more people, not only at the graduate level but also at the undergraduate level, in order to have an important critical mass to be able to face the challenges that we are living right now and those that are to come". He also recommends that primary and secondary school children, as well as high school youth, be taught about programming, not with the goal of making them computer literate, but so that they have at least some idea of how technology and artificial intelligence will work in their daily lives.

Conacyt will build an artificial intelligence consortium

With the aim of joining scientific and technological forces in the field of artificial intelligence, six research centers belonging to the National Council for Science and Technology (Conacyt) will form the Artificial Intelligence Consortium. Dr. Víctor Rivero Mercado, general director of the Center for Mathematics Research (Cimat) and coordinator of the consortium, explains that the consortium seeks to have a single instance in which artificial intelligence capabilities are located in Mexico.

"We want to structure and bring together all that we already have in research and development of artificial intelligence, as well as the mathematical models that are basic tools, infrastructure (laboratories and machinery) and other elements necessary to be able to establish collaborations and thus be able to advance more quickly". The 68 specialists (researchers and members of Conacyt Chairs) who will make up this consortium know that they cannot focus on everything and have decided to invest heavily in artificial intelligence applied to mobility, medicine, and public policy.

What about ethics?

In addition to highly trained human resources, research, technological developments, and infrastructure, something that is basic is a regulatory and ethical framework for healthy interaction between humans and intelligent machines. As explained in this text, artificial intelligence can know our darkest secrets and can use them at any time at its convenience. Although it is in the weak phase, artificial intelligence already has a great impact on daily life and it is predicted that in the coming years this impact will be even greater. In view of this scenario, the ethical and legal rules to be used are already being established.

Regarding the ethical issues, the specialists interviewed for this report agreed that it will not be possible to teach the machine an ethical framework, what will determine the application will be the people, the engineers, and programmers, who must be clear that they must build intelligent machines at the service of the human being and for his benefit. "If the computer is going to make ethical decisions, it will be the ones that are programmed, then it will be the programmer's ethics. We must remember that machines are our slaves because the programmer tells them what to do, the machine has no reasoning or free will, that is, it is not making decisions on its own," stresses Raúl Rojas.

As for the legal side, Dr. Jesús Manuel Niebla Zatarain, an expert in legal computing, explains that the development of artificial intelligence has forced different international jurisdictions to consider measures to regulate the operation of this type of device. "These guidelines establish, based on the laws, the way in which intelligent devices should operate. (So that) developers must provide these devices the ability to work as established by law," said the professor and researcher at the Faculty of Law Mazatlan, the Autonomous University of Sinaloa (UAS).

He stressed that currently artificial intelligence is not fully regulated, but efforts are underway in several countries, including Mexico, to ensure the ethical use of this technology. In fact, in Mexico, the Escuela Libre de Derecho, the Institute of Legal Research (IIJ) of the National Autonomous University of Mexico (UNAM), the Centre for Research and Innovation in Information and Communication Technologies (Infotec), and the UAS are working on legal and ethical proposals with a preventive approach, which can be added to the national strategy on artificial intelligence.

In 2063, machines could reach the intelligence of humans

All that has been explained so far has been weak artificial intelligence, that is, the most basic, which is the level that artificial intelligence currently has, but for 2063 you could have machines with the level of human intelligence. This is pointed out in the article When will AI exceed human performance? Evidence from AI Experts, by researchers from Oxford and Yale Universities, published in the Journal of Artificial Intelligence Research - which compiles the work and a survey of 352 AI experts from around the world.

"Researchers believe there is a 50 percent chance that artificial intelligence will outperform humans in all tasks in 45 years, and automation of all work in 120 years," the text explains. When that level is reached by a computer, there will be the talk of general intelligence (GI) or strong intelligence, and the machine will have the ability to understand thoughts, motives, intentions, and expectations, as well as interact socially.

In the year 2063, Rafael Pérez y Pérez hopes that Mexica has evolved enough to write interesting novels and not just short stories, as it does now. These stories will not be developed in the style of some human writer, by then, Mexica should already be a consolidated "writer" with a style of his own that will most likely conquer both humans and other intelligent machines.

What is artificial intelligence?

Artificial intelligence is an interdisciplinary field that has historically taken over knowledge and research from philosophy, mathematics, economics, neuroscience, psychology, computer science, cybernetics, and linguistics. Realization of intellectual properties in computer systems.

Artificial intelligence is a sub-branch of computer science that deals with the automation of intellectual behavior. Artificial intelligence is also defined as research on how to make computers do things that people do better today, or research on computing processes that allows them to perceive, judge, and act.

The term "artificial intelligence" (AI) was coined in 1956 by John McCarthy. Artificial intelligence is believed to be the study of the formal properties of problems and methods of solving them, and the part of computer science that studies symbolic, non-algorithmic reasoning processes and the representation of symbolic knowledge for use in computer system intelligence.

This work whose author is Verenise Sanchez, Mexico City, via the Agencia Informativa Conacyt,is under a Creative Commons Attribution 4.0 International license.