It will not be wrong to say that Artificial Intelligence is the future of the world. Everything around us is becoming more autonomous and more automated. It will not be long before most of the operations around us will be run by this technology, as it is dominating almost every sector already.
This innovative method has been used earlier to play games like chess and has received tremendous results and feedback. Recently, the company, DeepMind has launched a system that will allow the AI to work in a playroom setting on its own without human assistance.
They formed a gaming area similar to a playland called the XLand with multiple games just like they are present in a real playland. The platform had different games and activities. The system was not fed with information about each and every one of the games. It was left to work on its own and devise solutions on the go by creating new algorithms and solutions. When the results were observed, they proved to be a success as the system could generate solutions even for new and unique situations it was exposed to.
This has opened multiple avenues for the application of Artificial Intelligence. It states that there is no longer a requirement of training the system with tons of data before it can be made to operate. The basic algorithms and training can be provided, and the rest can be handled well by the system itself. It is a huge leap in the field of reinforcement learning.
It was made possible by exposing the system to dynamic tasks and then Population Based Training (PBT) was used to adjust the situations it was put in. frequent training runs were linked with one another so the new one could still use information from the older run. The study is being called the “Open-Ended Learning Leads to Generally Capable Agents”.