Intuitive AI is an experiment that is being conducted to generate intuitive behavior in Artificial Intelligence and is still ongoing. In this official documentation, we are going to see how everything works step by step and also what are the future possible enhancements.
This study is all about finding:
It's a big statement to make but that is the motto this study revolves around. Here is our abstract:
Intuition does not have a specific definition because no one can define what exactly it is. An easier way to understand it is when we compare it with intelligence. Intuition is based on feelings, while intelligence is based on logic. Intuition has more to do with a gut feeling, rather than a calculated decision-making process. If we can define anything in a mathematical model we can make computers do it. So, to define intuition on a mathematical model we need to see what causes it and due to intuition’s heavy subjective nature, it is not easy. So, we placed some real-world psychological principles that influence intuitive decision-making, into a simulator and conducted focus group sessions while collecting data of every game played, with a vision that these intuitive patterns which are coming from a human brain can be used to define a mathematical model for intuition. We believe that with this data that the simulator is generating we can motivate further research into intuitive AI.
When the computer language Java was first introduced it became the base of every single modern programming language and what they did was basically define the real world in terms of a mathematical format. We have inheritance in our real life and Java also introduced the concept. Though it was different in nature but in functionality it was the same. SO, if you are a java developer you can relate to this example and this is one of the many motivating factors in developing this simulator based on the real world principles that derive the intuitive decision making process in our brain. For computers things might be a bit different but eventually they'll be able to take non-calculated risks and when they do then they need to work hard to make that decision turns out to be right.
This documentation is structured as:
Understanding Simulator In this section, we'll discuss in detail how the stimulator was developed what is the logic and rules behind it.
Understanding Dataset A brief guide as to how the gameplay data is stored and what each column of the data sheet represents.
Exploring Generated Data Data that is being generated on every gameplay is being stored in a particular manner. So in this section, we are going to have a look at how the generated data looks like. Performing Exploratory Data Analysis (EDA) on the generated data. Putting it in shape and ensure that it is ready to be fed to a machine learning model.
Experiments with DataSet We fed the data to a neural network with a hypothesis that this neural network should never stop because it'll never improve or get better accuracy. We tried it using different machine learning algorithms and changed values of variables to see which makes it work. This and some other interesting experiments we did with the data are all mentioned in this section.
Future Enhancements What is the future of this project and how can you advance this research.
Have anything to share or ask? We have our community where you can share ideas, ask questions, or show something new that you did. Access our discussion community here: https://github.com/ibjects/IntuitiveAI/discussions
IntuitiveAI is an open-source project and our official repository can be accessed here: https://github.com/ibjects/IntuitiveAI
Here is the link to the IntuitiveAI simulator on the web: https://intuitive-ai.web.app/
Here is the link to CoLab Notebook where the analysis is currently in process: https://colab.research.google.com/drive/1rwPrzAS6J2lW0RVjqGaLpm4KaPwGGVdi?usp=sharing