The potential of these devices generates speculation that ranges from "magic" or "science fiction" to tangible reality, as well as ethical discussions, four experts in the field told the newspaper.
Before going down this road, it is worth appealing to definitions. The analyst Carlos Álvarez, who works at the company Idatha, recognizes that it is a "complex subject to address". "When we talk about AI we refer to a machine that can apply cognitive functions similar to those of a human", from the introduction of complex algorithms, which allow them to carry out those requested behaviors. "A simple definition of AI is: intelligence put by machines," says engineer Gastón Milano, Genexus' development manager, who adds that applications of these characteristics have been present since the 1970s. For Álvarez, AI "is a living science", because "today AI includes knowledge disciplines that did not exist five years ago".
As in other technology sectors, the entry of "big players" - IBM, Amazon, Google, Microsoft, SAP - into the field of AI has facilitated access to platforms that allow local developers to "walk on them with their applications. Engineer Benjamín Machí founded Idatha with four colleagues less than four years ago, and since then, he's seen accelerated changes in that sector. "One of the things that have changed is the availability of platforms that apply to this type of technologies that allow people without much knowledge of programming to generate their applications, obviously with their limitations". "Access to these technologies is increasingly open and available to everyone," he added.
For his part, the engineer Eugenio Garcia, product manager of Genexus, explained that the Uruguayan creators of AI applications "we integrate to platforms that already exist".
Carlos Álvarez: Three Vectors
"A perfect storm has formed, between the advance of AI, the advance of what is called 'internet of things', in which we are full of visual sensors, speed, all kinds, which are providing data of what is happening in the world, and the computational capacity that is in the cloud. These three aspects make it possible to do certain things that were impossible before, such as companies with two people starting an idea as a seed and can revolutionize different business domains".
For García, both "private" and "academia" should bet on training people to develop this knowledge. "We have the capabilities to make use of the platforms and take them to concrete things", but "there are very few people preparing to develop algorithms on their own". "There is a lack of training and academic offerings in the different faculties, and AI is an area that will continue to grow within information technologies."
The applications
The platforms available in the market allowed local companies to advance in different areas: "machine learning, deep learning, natural language processing that allows machines to understand what humans experience, and computer vision, which allows machines to understand what appears in images and videos. Progress has also been made in voice recognition, which allows a machine to listen and respond," Alvarez said.
Gastón Milano, engineer and research and development manager at Genexus, explained that clients "ask us to be one step ahead, foreseeing what is going to happen. They ask for a technological insurance, that we provide solutions to real problems that, over time, are overcome in different ways. "Today AI is a way to solve old problems in new ways." As an example, he pointed out that Genexus develops an application for a metallurgical company that sources scrap metal and was being harmed by the employee in charge of quoting it. "We developed a tool to make a recognition of the scrap metal they buy through photographs. An artificial intelligence engine is trained with many photographs, telling it how many elements - copper, lead - are in that scrap, and then it is provided with a new photo telling it the percentages of each one of them". Subsequently, the system itself is responsible for setting the value of that merchandise.
Álvarez, meanwhile, explained the functioning of applications which provide information to commercial brands and political leaders about what is said about them in social networks. "From the information obtained, we try to understand what is said, what are the objections, what are the things in favor; that is a basic analysis. After that, brands and politicians try to know what kind of people they are or what areas they are from, and that is extremely complex." "We can help understand what things are going on, but in no way with these kinds of tools can we tell them that if they say a certain thing or behave in a certain way, a certain thing will happen that will lead people to vote for them." In this case, the AI tool "is an input, just like a poll can be," he said. The tools which permit the reading of images are used "to read faces and detect if a person is sad, and also to analyze clinical images and detect skin cancer, for example. The technology is the same, although the application changes". "You can study traffic, people's driving behavior or understand how the behavior of certain animals is affected. All the information is analyzed and processed, there are people working in different areas and more and more are applied to improve results and understand what is happening", added Machín.
Myth and reality
Those who work in AI must establish a limit "between what is real, what we have, what is projected and what does not really exist". "We cannot detect all the failures that a machine can have, because, a priori, we do not know that universe of possibilities and we cannot predict all the unforeseen events". "From that magic behind the belief that everything can be done through AI, applying algorithms, we are quite far away," warned Alvarez.
However, it is clear that "exponential advances" in AI will mean that some activities now performed by humans will be done by machines, leading to the disappearance of jobs. Examples abound on all sides to demonstrate this assumption. "Any work that is automatic and methodical today is at risk," Milano stresses. "We have to know how to change, and the basis is education. Can we have autonomous people in agriculture, doing things that today a person does? The answer is yes. But if you ask me: can I use AI to make an entertaining television program? I don't know, because that's where you have to put your head".
Gaston Milano: Knowing the domain
"To solve a problem, the last thing you need to be is a mathematician, because there are ready-made programs and we can use them. So, today, rather than a game of engineering, it is a game of combinations of knowledge that are available. So, in a team there are engineers who know about AI, but there are also common programmers who know how to integrate services, and suddenly a young person who has completed the Ceibal Plan's Jóvenes a Programar can participate in these experiences without knowing how AI is made. And the most important thing of all is the person who knows the domain in which the solution is going to be implemented, which can be a technician or an agricultural engineer if we talk about topics related to the field, or a doctor when we are researching radiological images".
Creative activities or those involving the staging of "values" are still far from being solved by machines. Although technology can identify cancerous tumors, "if you ask an AI tool to cure that disease, it might decide to kill us all, because what it should do is eliminate the cancer, but I didn't give it the value of life, which is something intangible". "More jobs should be created in creative areas, in the invention of things that AI can hardly do, because its limits are in the emotional, in the creative," he said.
Currently, AI models "know a very small world, which is what we give them, which we program". "When more advanced models appear, that can interact, that know that there are other beings apart from them, then we will be talking about another evolution, another type of machine, that would have emotions, but it seems to me that we are far away from that". "In the end, they are machines; we are the ones who created them, and the result of what happens with them will be the responsibility of our morals and ethics, both in their use, as well as in what we do as a society, because we cannot leave everyone out of the labor market, since we would generate much bigger problems than those we are trying to solve," García concluded.
Goal to the future
If advances in sensory technology can detect the expressions on human faces, why haven't sensors yet been placed in soccer goals to put an end to arguments over whether or not a ball crossed the goal line, the daily asked Milano. "Actually that exists, because there are sensors for everything you can imagine. In this case, it would be something very similar to the 'hawk eye' used in tennis. But in this issue enters the culture of sport and the magic that would be lost in football if such devices are used," he replied.
Within these limits established by culture, Milano adds "the habit of working with averages". "In the case of viticulture, we calculate how much will be the average of grapes that we will obtain for the production of wine in a certain territory. What is customary is to take ten samples and calculate an average. But now we can actually take pictures of the whole vineyard and calculate, grape by grape, exactly how much we will be able to obtain". Likewise, "now, when there are discussions about how many people attend a demonstration or an event, I'm sure you can count the number of people who attend". "Let's not argue anymore, let's do it," he proposed.
Tesla: the fantastic car
A couple of self-driving cars are driving around the streets of Montevideo. Orlando Dovat, president of Zonamerica, purchased one of the first Tesla vehicles to arrive in Uruguay, as he posted on his Twitter account in November. The vehicle, model S P90D, has an autonomy of charge in its electric batteries that allows it to travel 500 kilometers (310 miles). The Tesla can be started remotely via a cell phone application. When in motion, the car respects the destinations mapped out, as it does with the distances that separate it from other vehicles traveling on that road and with traffic signs. It also parks without human assistance, either parallel or perpendicular to the roadway. When allowed, it can reach a speed of 110 kilometers per hour, although it has enough power to go well beyond that range. Travelling in one of them becomes an experience difficult to forget.
Source: La Diaria
Connect