Recently Google has announced its New AI which is based on a Machine learning framework so that the game developers can use that system to train the agents for game testing efficiently and quickly which will help human testers to keep their major focus on complicated problems and bugs.
The resulting AI does not require an expert in machine learning and it also works with a wide range of popular game developers of all genres. It can also train a Machine Learning policy that helps in generating the actions of the game from the game itself in just one game which takes time of less than an hour. For using this AI the Google has provided an open-source library that shows how to use these techniques.
Playing video games simply is the most basic form of game testing. There are so many serious bugs in a game such as falling out of the league or crashing and they get detected easily and fixed easily. The developers use to face a major challenge to find the bug in the vast state space of all the modern world games.
For the solution to their problems and reducing the time to check the crashes, the developers focus on to train system that just plays the game simply.
This is an effective way of solving the traditional problems in the modern world which is that enabling the developers to train an AI which consists of a group of game testing agents which is better than training an individual which is proved to be a super-effective idea that agent is playing the whole game from end to end and every agent will perform a task individually for few minutes which is referred as “Gameplay loops”.
Google AI ML, The system will be bridging all the gaps between the virtual imaginary centric world of the video game and the data-centric world of machine learning which is one the most faced obstacle in applying the ML during game development. The high level of semantic API is easy to use and allows the system to adapt to the specific game which is being developed.