On March 15, 2016 may go down in history as a key date for artificial intelligence. That day, the computer developed by Google, AlphaGo and South Korean Lee Se-Dol, ended his series of five games to play go, a form of great tradition in Asia and whose difficulty, they say, is much higher than that of chess including for a machine. The end result was incontestable: 4-1 AlphaGo defeated its human opponent, the first time a computer defeats in the history of this game.A milestone that demonstrates the level of development achieved artificial intelligence. A technology in everyday life and a basic level, can be seen, for example, websites that recommend purchase certain products based on purchases made before, and that sooner than later, hit the road as autonomous driving cars. However, its application goes beyond that.
Machine learning (machine learning, in its English translation), is a branch of artificial intelligence that lets the computer, through complex algorithms to process huge amounts of data, learn from them, and draw conclusions as, for example, predictions: "to recommend content that is relevant to the person, optimize logistics systems if you have to order new products, you can predict demand ... in terms of industries, almost all may be used, if it is already using" says Nuria Oliver, scientific director at Telefonica I + D. "Thanks to the existence of big data, and capacity ever greater computing, are taking very great strides in the field of artificial intelligence, because they are algorithms that need data to learn, and a lot of computing. We already have, and now we are becoming witnessed major milestones in this area. It is a hot area. "
Oliver was one of the invited speakers Connect Papis Congress, which met in Valencia experts and professionals in the field of artificial intelligence. There, the pre-series platform, owned by Telefonica, Big Ml, introduced a machine capable of selecting between five startups, which one has a better chance of success and therefore what is best for investment. "He has access to a wealth of information that a human would be very difficult process," says the president of preproduction, Francisco Jose Martin. But can a reach strategic decisions, for example, in a company machine? "How many managers in Spain can reach the level that can have a player go?" Martinez asked. "Few, sure. If an algorithm can beat a human in this game, imagine how well they can make decisions, "continues the president of preproduction, which limits, yes, the fields in which a machine can sufficing alone:" If the decision is to launch a marketing campaign, or implement a product, are things that someone will have to define and work. But matters that are automatable, like putting a price or make a purchase order are tasks that a machine can do. And it is very easy to measure the impact of their decisions. "
Nuria Oliver believes, "depending on the complexity of the decision and the impact that will have in the real world, we have situations where it is fully delegated decision making to an algorithm. There will come a case where programs make decisions because they are things that can automatizeize "However, bet on a hybrid system, in which the algorithm is the to analyze the data and present the findings, and is the human who has the last word. Oliver argues that, still, a person can bring skills that a machine does not have. "It will be increasingly necessary symbiosis between algorithms to help us take advantage and understand the data, and humans, which can contain a notion of context, or non-numerical factors. In short, that artificial intelligence is another input in decision-making. "
For its part, the director of the Institute of Artificial Intelligence Research CSIC, Ramon Lopez de Mántaras, machine learning emphasizes its predictive utility, which can be appreciated in sectors such as retail, "The possibilities are immense. From the business point of view, to maximize benefits to predict pandemics at government level ". Detailed that even a small company could apply this technology because they "do not have to invest in sophisticated hardware or software almost. Yes in people who have some knowledge of artificial intelligence to exploit its full potential. " Indeed, with respect to human role, Lopez de Mántaras is more blunt: "When is economically and socially damaging to run a standalone software, do not allow it. I'm against giving full autonomy to machines. " For example, in a human resources department. "They are using programs that filter out 90% of resumes, and the other 10% and analyzes a person. I think it's a huge mistake because you're focusing on the future of people. "