Glossary of terms

Agent Environments: where the Agents interact with the game world. Agents can observe the game and produce a subsequent action. These environments are where the training takes place.

Deep Reinforcement Learning: A subfield of machine learning which in turn is a subfield of Artificial Intelligence, Deep Reinforcement Learning is the science of combining both deep and reinforcement learning so machines can make decisions, learn from their actions and mistakes (similar to how a human would), and improve itself.

Game balancing:

Area of game design that offsets the strongest traits of a character or strategy with an equally strong drawback in a specific area so as to avoid character or game approach dominance. To create a positive gameplay experience, game designers typically tune the balance of a game iteratively:

  • Stress-test through thousands of play-testing sessions from test users
  • Incorporate feedback and re-design the game
  • Repeat 1 & 2 until both the play-testers and game designers are satisfied

This process is not only time-consuming but also imperfect — the more complex the game, the easier it is for subtle flaws to slip through the cracks. When games often have many different roles that can be played, with dozens of interconnecting skills, it makes it all the more difficult to hit the right balance.

Gamer Technique: The style the player employs to interact with a game and what makes them enjoy it.

Gameplay Environments: Gameplay environments of the simulated world carry out the game logic with actions performed by the Agents, producing the next state.

Human-like Avatars: Human-like Avatars are a simulation created by the Meta-Agent to play specific game levels.

Meta-Agents: A Meta-Agent is a complex set of Neural Networks with a certain level of intelligence that are trained by our GameAI team based on each client’s unique needs and game title. Meta-Agents are constantly learning new mechanics and their intelligence grows simultaneously. Meta-Agents produce Human-like Avatars to play game levels.

Playtesting: Quality-oriented gaming process that helps game designers test a new game or level to discover new bugs or flaws before the game or new level is released to the public. Traditionally, playtesting is performed by human testers to try out the video game and provide feedback. Thanks to smart technology, namely Deep Reinforcement Learning, automated playtesting through artificial intelligence agents that play the game with a specific goal determined by the game designer.

Can’t find the answer to your question?
Contact us and we’ll get back to you as soon as we can.
By submitting a form via this website, you consent that your contact can be used for the purpose of contacting you, responding to your query or directing you to the appropriate person within our company.