After a busy week 8 of catching up with work progress, we have found a more comfortable pace and learned more about the team and our clients to move ahead. I will share more details about what I mean.
Game Design Updates
Focus on “light magic”
Despite the adaptive nature of how our game allows players to focus on the “light magic” or “dark magic” directions, we have decided to focus on the light magic direction so that we do not confuse our players.
The “light magic” direction entails players to generate statements which is viewed as toxic by the AI models but the players think otherwise. This trains the users to think in a non-toxic manner and help to sieve out preconceived biases within the AI model.
Part 1 card creation process
We have created a new wireframe to showcase the entire card collection process. The overall concept is similar to what is shared last week.
By default, the player will generate a random sentence on their own which look like this:
If the player requires external help, they can click on the “Get Inspirations” button where 3 words will be randomly generated to help the players form these biased statements, looking like this:
On top of that, if the player still require help, they can click on the select mode where 3 words each for adjective, verb and noun will be created where players can form a sentence by dragging and dropping as shown below:
Lastly, the most important step, the player will receive a feedback from the AI model and eventually make a decision on forming a card or re-creating a new statement to try.
Background linking back to the storyline
We have brainstormed many ideas to come up with a suitable background similar to be used like conveyor belt like factory machine, school or classroom settings for robots to learn and something similar to the baby boss concept. Each has it’s own pros and cons and we have yet to find a really good direction at this stage. We will work on brainstorming this further over the weekends and generate some ideas to be shared with our clients on Monday.
This week, we have successfully create the part 1 which is the card creation of our game. This allows players to generate statements where AI thinks it is toxic and the players think it is not toxic and eventually create a card based on the card generation.
After connecting to the Firebase database, we have also successfully enhanced the authentication flow setup from our gameplay to Firebase. The created card in part 1 will also be saved directly to the database for future use.
Hugging interface update
We have also set up a pro account from our game to link directly to the hugging interface online where the AI model is hosted to be used. You can check the model out here.
After a few rounds of iteration with our clients, we have finally decided on the(90%) confirmed storyboard. Our artist has started working on animation to bring it to life.
Upcoming part 1 background design
With the storyboard and part 1 wireframe confirmed, we started brainstorming on ideas to incorporate a background to the storyline. The artists are working on it and it is planned to be shared with the team and our client on Monday.
Playtest Updates (Internal)
We started our first internal playtest on the WebGL build this week. Here’s the link to play it. We have successfully managed to deal a fixed set of deck to each player. With each round, the players have a fixed time to place a card based on a random topic and vote on a card that relates to the topic the most. Feel free to check it out (even though it requires at least 4 players to play at the moment…)
Positive Playtest Feedbacks
The current gameplay has a neat interface and clear instructions with a goal for the player which makes it easy to understand.
Things to work on Playtest Feedbacks
The interest curve is still fairly flat and players may not have enough context both from the generic card and topic formed. More work may need to be done there to enhance the player’s experience.
In conclusion, we are working on these areas concurrently while the tech team focus on the development of the game.
New Experience with our Clients
As we finished conceptualizing and moving closer to production, we are having more back and forth discussions with our clients to learn more about their opinions to carry on with our work.
In this week, we learned that we tend to put more emphasis on the game play aspect of our product while our clients are focusing more on the practical usage of the product being delivered. This made us realize the importance of discussion. In order to create a Game with a Purpose (GWAP), we really need to find a careful balance between gameplay and work in our product.
In retrospective, it is not easy but we are seeing the benefits of this collaborative work and we have to create a final product which our clients and us can be proud of.
AIAuditors Production Update
Currently we have decided not to go for external playtests at the moment as the players will certainly lack context to understand the game. We are working very hard to try to incorporate the context to be used for our upcoming on the playtest day in week 10 Saturday.
With our part 2 complete and part 1 going to be completed by Monday, we are on track to combine these 2 parts together in week 10 to be used for our upcoming playtest session.