So-called reinforcement learning is a technique of training artificial intelligence, where the algorithm is rewarded for making correct decisions and being punished for making wrong decisions based on the results obtained earlier. According to this idea, with the right amount of time and effort, the system finally sets itself the best course of action for itself. For some time it has been a common practice in training artificial intelligence. She recently used Google’s DeepMind, for example. With her help she taught artificial intelligence how to successfully overcome obstacle course.
It is clear that the artificial intelligence from DeepMind uses creative solutions to overcome the obstacles that are on its way. However, it is worth noting that most of the time these solutions do not look natural, although they provide the best results. This shows how potential future artificial intelligence can be – since robots in terms of mobility will not necessarily be as limited as humans.