The creation of human-like thinking machines is one of the most frequent dreams in science fiction. There are many works of the seventh art that have imagined the existence of computers or smart robots in the future, for example, the film Artificial Intelligence directed by Steven Spielberg, in 2001.
Humans and robots
Many researchers agree that, at some point, we will live with machines or robots that will make our lives much easier and will replace humans in many more tasks than today. However, the advances of Artificial Intelligence happen slowly, while the scientific debate about how robots can match humans and even surpass our capabilities remains open.
But, even if it seems incredible, real science is increasingly approaching science fiction. A team of scientists from the New York University, Toronto University and the MIT (Massachusetts Institute of Technology, Boston) has managed to develop the first algorithm for an Artificial Intelligence or Robot to learn new concepts similarly to a human.
Learning for robots
The leader of the group Brendan Lake and his colleagues have created a program capable of learning to write characters from an alphabet from a sample. Today, some robots already learn as humans, take an action from a model or example and then repeat it. However, the difference is that while a person needs a few examples, Artificial Intelligence needs thousands of examples to become familiar with a task and then be able to repeat it.
Nevertheless, Brendan Lake, lead author of the study, stresses that humans are still much better than machines at the time of learning. But he argues that his investigative work could be a step in the future to narrow the machine learning gap and develop new methods that allow machines to learn to do other tasks.
There is no doubt that machines are increasing their powers more and more, as technology continues to advance exponentially.