MAI (Mohan Artificial Intelligence)

[ravimohan1991]

Here we chalk out the blueprint of the Game AI that we intend to embed into Karma. We do that by asking a series of several relevant questions about Game AI and attempting to answer them.

Certainly one of the ways to rig Karma with Game AI is to use already existing libraries, which have survived the scrutiny of time and are (partly?) successful.

We hope that the Game AI framework will reveal itself as we build up the Game World harnessing the complexities of programming, art, and gaming community (developers and players). This way it will be more in sync with the real and necessary needs of the community and thus fulfil its purpose.

Questions and Answers

Q. What is Game AI?
A. Game AI is the companion of the human player.

Q. What are its objectives?
A. The following:

• Bring appropriate complexity to the gameplay.
• Be very explicit (no manipulations) and principled.
• Ability to have independence and self reliance. In other words, ability to make choice with accountability.
  Note: Choice should be more likely based on principles and ethics rather that myopic outcomes.
• Awareness, survival instincts (not like virus though), learning, curiosity, empathy, pride (not egoistic), faith, vulnerability.
• Keep the player engaged not by trick or illusion, but by sincere efforts.

Q. How do we implement Game AI in the abstract world (Game World) based on the computations done in the materialistic world and abstract world.
A. We will start with baby steps. For example right now Karma is able to load models with basic camera and movement. We will see if we can effectively embed awareness at this stage.

Q. Does Game AI need to be curious/enthusiastic about the way Game World is created?
A. This is an interesting venture to explore. We will be entering into the domain of recursion scheme and closer to the definition of consciousnesses.

This list is by no means concrete and complete. One of the aims of these discussions is to achieve just that.

We will keep in mind that these questions will certainly help us to define an intelligence which will be expressed given the constraints of programming (not necessarily art) and that too on a Turing Machine. The aim is to achieve the highest possible fidelity.

We look forward to reading your opinions and contributions related to the subject. If you know any references which are exploring the similar uncharted routes or you feel that a question is unjust/not clear, feel free to write a word or two.

The branch where we will implement these ideas is MAI.