On the cover

artificial intelligence (AI) will boost productivity, make the world more efficient and safe, extend life expectancy, serve to predict the future, prevent catastrophes and even combat climate change. The companies that develop this new era based on algorithms and big data present it as the holy grail and tiptoe through the obvious dangers that it also entails. Because, according to different studies and analysts, between 14% and 40% of current jobs risk disappearing due to the combined effect of these systems and robotics.

“Human beings will be forced to continually train so as not to become obsolete,” advances Robin Li, Dubai’s CEO. But there is a scenario in which even such continuous training would not be enough to guarantee jobs: that of technological uniqueness. While still a more appropriate concept for science fiction stories, this hypothesis draws a future in which technological advances lead to a super intelligence that far exceeds that of the human being. And there is no shortage of scientists who consider it a less remote possibility than many others want to believe. Could the future outweigh fiction? Especially in China, the country that has set out to lead the development of artificial intelligence: it is already the one that invests the most in the sector, the one that has the most public institutions investigating and the one with the most patents: 57% of the total, compared to 13% of the US and 7% of the European Union, according to the 2018 Artificial Intelligence report, to the European Perspective,of the European Commission. And it has 17 of the 20 most relevant research institutions in the field of AI, according to a study by the World Intellectual Property Organization. “While I would like america to win the AI race, if I had to bet I would do it for China,” says Thomas H. Davenport, author of The AI Advantage. “China has many advantages: a particular government, an inexhaustible source of money, a growing number of smart scientists, and a huge population that loves digital,” he argues.

It is also the Could the future outweigh fiction? Especially in China, the country that has set out to lead the development of artificial intelligence: it is already the one that invests the most in the sector, the one that has the most public institutions investigating and the one with the most patents: 57% of the total, compared to 13% of the US and 7% of the European Union, according to the 2018 Artificial Intelligence report, to the European Perspective,of the European Commission. And it has 17 of the 20 most relevant research institutions in the field of AI, according to a study by the World Intellectual Property Organization. “While I would like america to win the AI race, if I had to bet I would do it for China,” says Thomas H. Davenport, author of The AI Advantage. “China has many advantages: a particular government, an inexhaustible source of money, a growing number of smart scientists, and a huge population that loves digital,” he argues.

It is also the We have always thought that human intelligence was impossible to copy, and now we know that one day we will be able to understand it well enough to replicate it. That assumes it’s not irreplaceable,” says one scientist. A glimpses of super intelligence?

“This super intelligence is coming and we may not even realize it’s coming,” adds Brian Suburban, director of the Auto-ID Lab at the Massachusetts Institute of Technology (MIT) [interviewed by Jaime Susanna at Become, the Iberia-American Strategic Communication Summit].

And it gives examples of some steps that are already being taken in that direction. “Image recognition over the past seven years has already surpassed that of human beings. That’s why, in immigration controls, cameras can recognize faces better than agents. It is a technology that is also used in medicine. In the case of retinal tumors it has been discovered that the computer knows how to distinguish whether a retina is male or female, something no doctor can do. That would be an example of super intelligence in which the computer sees things that escape the human eye.”

On the other hand, computing power grows exponentially while chips get smaller. Peter Abbe, a UC Berkeley professor and chief scientist at Embodied Intelligence, agrees: “There is still no computer power machine to reproduce the human brain, but it can be achieved in the cloud, with a network of computers.”

In Abel’s opinion, that’s not the biggest problem. “As the chips progress, they will also get cheaper. A machine with a person’s computing capacity could even cost less than the minimum wage,” he says. José Corrosion, Professor of Computer Science and Artificial Intelligence at the Autonomous University of Madrid and senior researcher at the Institute of Knowledge Engineering (IIC), coincides. “Some argue that quantum microprocessors are difficult to control and that Moore’s Law, which says that every two years the computational power is doubled, will prove wrong because the chips will face insolvable physical barriers. I, however, believe that new technological proposals will be presented and I see it feasible that the computing capacity of the brain can be replicated in a reasonable time.”

Not for nothing, the European Union’s Human Brain Project is also looking for something similar, but from a broader perspective, with the Brain Simulation Platform. While Abbe believes that these advances may be the first step on the road to super intelligence, Corrosion does not believe that it will become the germ of technological singularity. “A machine can do a lot of calculations at a huge speed, but it needs a conceptual framework to go further. We have to understand the brain also from a physiological point of view.”

Similar lines show Ramon Lopez DE Tarantulas, director of the Research Institute in Artificial Intelligence of CSIC and one of the scientists who see the concept of super intelligence with more skepticism: “Computational capacity is compared to the electrical activity of neurons, but in the brain there are more glacial cells than neurons. Now we know they play an important role, but not how to reproduce them.” Lopez DE Tarantulas says that it will be possible to create a neural network comparable to the brain in terms of its number of process units, but warns that only electrical activity will be being modeled, forgetting chemistry, which is essential for processing information. “I mean, in no case will it be a brain,” he says.

Leave a Reply

Your email address will not be published. Required fields are marked *