Reinforcement Learning & an AI that can build another AI:
AI is one of the fast-paced and well-developed fields of experimentation in today’s times. Artificial intelligence as the name suggests is the kind of intelligence that is developed in the computers to let the make them perform tasks just as the humans. These functions might include speech recognition, translation, decision making and much more. There are various parts of Artificial Intelligence, and one of them is Reinforcement learning.
Reinforcement learning is becoming increasingly developed and might be the future of the AI robots in the world. Let us get to know a little more about Reinforcement Learning.
Reinforcement Learning In A Nutshell :
Reinforced learning is one of the most researched components of AI. It involves teaching the computer to learn something with the process of giving out various rewards and punishments. For example, we train a dog to perform a trick by telling him that he did right by giving him a treat and telling him he is wrong when he does not follow the order. This is a similar case of the Reinforcement learning in the computers. In the case of computers, it is not physical rewards. The computer learns to perform a task by the trial and error method just like in humans or other animals. So with reinforced learning, a robot can make its own decisions by learning from its bad experiences.
AI that has Built Anther AI:
Scientists have spoken about creating a robot that could build a new robot by itself. A few years back this looked like a distant dream, but now Google has made this possible. In May 2017, Google announced their new creation AutoML. The ML stands for machine learning. This is the Artificial intelligence that has created its baby artificial intelligence. The best part of the baby AI is that it is so advanced that it has outperformed its creator and also the scientists that have built it. This child AI is called the NASNet. The creation of a new AI takes a lot of time and work by various people to design it. Hence, Google made a unique approach and let the existing AI develop its own AI. It turns out that the process is a lot faster because the computer is coding way better and faster than the humans. The AI AutoML is now teaching the NASNet to perform various tasks and improving its functioning.
How are these two related?
The relation between the AI building another AI and the reinforcement learning is pretty evident. The child AI is being taught to perform various tasks by reinforcement learning. The reinforcement learning is the best way an AI can learn to perform human-like functions and can use the intelligence in the future situations. The thought behind the development if AI by and AI is that Google wants to give the smaller companies who do not have the required expertise to design their neural nets for their own needs. The AI and Reinforced learning are two very advanced technologies that have a bright future in creating the world a better place.