Idea Proposal

We propose a gamified problem-based learning platform to teach AI in a physical science class. The learning platform will specifically introduce classification concepts as well as ethics in the context of the subject. We are thinking of teaching kids about chemical reactions and ethical issues in a laboratory. We anticipate that as students interact with the game-inspired AI learning tool, they are expected to solve some tasks while in the process, they learn, for example how computers classify and promote their ethical reasoning toward AI.

Mixing two or more reactants forms a product creating a reaction or no reaction. There are some common physical changes that can indicate a chemical reaction has taken place. For example, a change in color or the production of an odor can be a result of the formation of new chemical species in the reaction. An increase or decrease in temperature can also be an indicator of a chemical reaction, as energy is often absorbed or released during the process. The evolution of a gas (such as the formation of bubbles) can also indicate a chemical reaction, as the production of a gas often suggests a change in the chemical composition of the reactants. The formation of a solid, also known as a precipitate, can also be a result of a chemical reaction. 

We are trying to develop a game in which we would predict Reaction/No Reaction of the product based on the reactants. One can see how accurate AI Bot is when comparing its predictions to the testing data that we already know the answers to. The activity would focus on training several different models. This can help explain why this model earned the accuracy score that it did. The features would be color Change, production of an odor, change of Temperature, evolution of a gas (formation of bubbles) or precipitate (formation of a solid).

We plan to teach a binary classification problem through our platform where the model will predict if a group of substances will react with others or not. Binary Classification (have 2 classes) - gives 0 if no reaction and gives 1 if there is reaction. We found an open source database called - ORD (Open Reaction Database) - https://docs.open-reaction-database.org/en/latest/# that we plan to use for training. We plan to build a game where it has some levels - starting from prediction between a few elements and then increasing the number of elements in higher levels and see how well the model is able to perform.

Proposed Activities

 Activity 1: Teaching students about Ethics in lab by asking to train students to train the AI model on what is acceptable to carry in a lab and what is not. Based on the training model the AI Bot will predict if other accessories can be allowed or not allowed to bring in the lab. 

Through this activity, students will be able to:

• learn about the ethics in a laboratory

• learn about why do we need a diverse a training data set

• learn about the importance of data and how it can lead to a good decision or a bad decision

 Activity 2: This activity is like a game in which we would predict Reaction/No Reaction of the product based on the reactants. The model will predict reaction or no reaction on the basis of features like gas formation and other properties.

Rough Sketch for Activity 2

After clicking submit button