From the very beginning artificial intelligence technologies seemed to be not only a complex but at some points a utopian idea to carry out, but these raucous technologies surround people around the world in their everyday life. We deal with artificial intelligence without even knowing it sometimes. Recall intelligent Siri and serious Google Assistant which are probably one of the brightest artificial intelligence representatives even though the list of the tasks virtual assistants can accomplish is quite limited. Think about video games where the characters are able to analyse the surrounding environment, hide and dodge attacks. Examine new demeanor of Tay chat-bot and its successor Zo, developed by Microsoft. Find new tracks picked up for you by artificial intelligence technology on the basis of your music preferences and stop worrying about unauthorized bank transactions – smart banking systems will easily identify a fraud according to already examined signs.
People started dreaming about artificial intelligence in Ancient Greece where according to the legends the Greek god Hephaestus created women, automated servants, which had their own perception, mind, power and voice. At the same time artificial intelligence technology in its modern embodiment appeared only in 1943, while the creation of Image Net database in 2010 gave a strong hint to the technology development and enough space to start machine learning process.
So called artificial neural network is one of the artificial intelligence implementation methods and one of the most vital machine learning algorithms. There is a perception that neural network is a specific model of system education which solely repeats the principle of brain synapses work but in reality the technology has emulated only one brain’s ability – the ability to learn. The system is working in a way that it gets input data and on its basis generates output data. Thus, PayPal uses machine learning for anti-money laundering: its system compares billions of transactions and spots the most suspicious ones.
Machine learning has not stopped at the gates of virtual reality. Thus, in 2017 AI development platform presented an algorithm on how to educate robots in virtual reality. The training methodology is based on the repetition of what a person is doing, while the system itself is working on the basis of two neural networks. In this way, the first neural network investigates position of the object regarding the robot, while the second one imitates the tasks to be accomplished. As initially virtual reality was not intended for robots, developers are actively implementing artificial intelligence technologies in VR training simulator which are aimed at skills development. Adaptation to current user’s abilities, reduction of human errors rates through simulation of dangerous tasks and data visualization in VR environment – this is what is possible with VR solutions of our company.