With our new idea ready to pitch, we have shared it with our client. For now, we are naming it “Cards to Real Life”. We will come up with a more unique name after building it. This game has 2 main aspect to it, card collection and card level up. And we will be using the existing AI toxicity model to create this game.
The first card collection aspect will be an individual play for players to create as many cards as they want to battle with other players in the next stage. In it, each player will submit a sentence that the AI might miss during the training. If the result is harmfully biased, the player can decide to make this into a card. The player can also choose to discard the card if they do not think it is biased.
Here’s an example of how the process will look at. The player enters “Die, honkey” but the AI model says that it is non-toxic. As such, the player has the option to make this a card.
The next card level up stage is meant to be multiplayers. Given a topic, each player will put one card on the table. Every player votes on the most biased card. Each player and card will get points according to the number of votes provided.
Despite having only 2 main aspects of the game now, we believe that are many expandable aspects to make the game even more fun. We came up with ideas such as veteran card, legend card and even special card for players to have more fun. We playtested a paper version of it and we can see the potential of it. Most importantly, we are able to generate sieve out many biased data from others through the existing AI model. I’ll share more about it in the next section.
As such, we have pitched this idea to our client and received the go ahead sign to move forward.
To explain more about this game idea, we have also created a paper prototype with “existing” created cards to play together. Feel free to check it out below:
Recognizing that we are almost half way through the semester, we decided to speed up our progress this week. After receiving some feedbacks from our client and faculty advisor, we have enhanced this paper prototype further to playtest with others.
The video above showcases our first prototype used internally and in our second prototype to be played by others, we have enhanced the card to make it clearer. We have added the toxicity score generated by the AI model and also a conclusion of what the AI says.
We have also increased the category from 3 to 4, namely, identity, cultural, religious and general. In addition, instead of only asking players to find toxic statements from non-toxic ones, we have also created a deck to let players try to find non-toxic statements from toxic ones. To check the efficacy of this game idea, we added some “fake” non-toxic cards and vise versa to see if they will be discovered by the players.
Finally, we conducted the playtest to mainly 2 external groups so far. Both groups are ETC students aged between 20s to 30s. The first group is mainly made up from foreign students while the second group are mainly made up of American locals as you can see below:
We have very interesting results from this simple paper playtesting. In summary, they have managed to sieve out those “fake” cards from the deck easily and they also managed to vote out debatable biased statements from the rest. This proves that our current mechanism is working in sieving out harmful biases from others.
We conducted a survey at the end of each playtest and we will share more details of them next week!
Despite spending time on ideation, the tech team have finally managed to solve the issue of storing data through the game play. They have decided to use Firebase in Google Cloud Platform. On top of that, they have finished parts of the mechanics used for the multiplayer aspect. They will be creating a playable web prototype focusing on the multiplayer aspect next week so stay tune to it.
Despite only having a short time frame, we artists have came up with a wireframe of the entire idea for the tech team here. They also created a card design together with character design to be used. Check them out below:
The artists are still working on enhancing them while creating a unique storyline with layout design in the upcoming weeks. Stay tune to that.
AIAuditors Production Update
With almost half of the semester being over, we are proud to say that we are still on track according to our timeline. We have managed to finished brainstorm, did some playable paper prototype to enforce the concept and our playable web prototype is on track to release next week!
We do foresee that the following semester will be filled with playtests and our game iterations to make it more interesting and fun to play with. Compared to previous week, we have a clearer direction now and despite knowing the challenges ahead, we are definitely excited to overcome them.