Introduction and overview of the project
Team Neo Security Lab, in cooperation with the CERT Division of the Software Engineering Institute at Carnegie Mellon University, is creating a web-based game testing its players’ degree of trust for AI systems. The game will be designed to entertain the player, while also collecting their choices to help with AI decision-making interface research in the future.
The goal for our team is to produce a purposeful game that generates large amounts of data regarding decisions and interfaces, without collecting any personal information. This game is about players’ attitudes towards AI systems, while keeping players anonymized. Neo Security Lab uses an application programming interface for AI decision support provided by the CERT Division. The game centers around those choices. This work will help AI researchers improve strategies for displays of AI-based decision support.
Team Members: Alan (Tianyi) Zhang Danny (Chien-Kuo) Kuo Hao LuYe Wu
Project Instructor: Mike Christel
Client: Rotem D. Guttman
What went well (or what went right)
Transfer from onsite to remote work part: Because of the influence of covid-19, we had to change from onsite work to work remotely, but our team completed a rapid and organized change in this process, and quickly invested in work within less than a week, even more efficiently.
Project design part: We work closely with client. At the beginning of the project, according to the needs of client, which is to collect the answers to 20 questions from users in a relatively short period of time and can’t let users be distracted or quit halfway, our project should prevent users from mechanical clicks. For example, some users will always choose a certain answer in order to complete in a short time. The most important aspect is that the client wants to import their own API to complete data collection and problem generation. For the needs of the client, our team quickly started research and finally selected a web platform to develop with. Because the clients want to collect data on various platforms and allow users to access easily, our client and instructor are very satisfied with our choice of Cocos2D for developing an html5 browser-based experience. During the project development process, we actively listened to feedback and suggestion from instructor and ETC faculty and communicated to client, And in a step-by-step iteration, we gradually confirmed that our project runs in a mode of interspersed with task questions, fragmented stories, and mini-games, because this model can maximize the efficiency of collecting data on multiple platforms, In terms of project design, the client is satisfied with the cooperation between our team and him, and we sincerely thank our project instructor for the help of our project and those heartfelt suggestions.
Mini game design part: For the development and genre of mini games, our team has actively made many versions of iterations and playtests, and listened to the suggestions of ETC faculty and project instructor with an open mind and finally won the affirmation of client. In the beginning of mini game development we made rough-like mini games, puzzle mini games, and action mini games prototypes that we brought to our client, In the end we chose puzzle mini game because minigame’s easy working with mouse or touch on phones, relationship to the story closely and our client’s preference for a patch panel like game. In the development stage of puzzle minigame, we first developed a simple version of the puzzle game of the link pipeline, after that, we got feedback from client and mentors. Client wanted to make the mini game more futuristic. Therefore, we turned the mini game’s link pipeline into a link laser and added a reflection effect. Before the soft open, we conducted many playtests, We had 34 sessions of data, including 23 questionaires following the experience. The loss of Playtest Day and the covid-19 situation negatively affected our playtest counts, but we were able to win the support of our client with our minigame iterations. And the testers agreed that the mini game added a lot to the entire project.
UI design part: In terms of UI design, On the one hand, our team has completed all the functions requested by the client, on the other hand our team wants to combine the background story of our project to give users a sense of the future and related to artificial intelligence. Therefore, under the guidance and suggestions of the project instructor, we
refer to the elements of Star Wars and the Matrix, and adopted a large number of blue and thick line design. In terms of fonts, we have listened to and collected Professor Shirley Yee ’s suggestions many times. In each version and weekly update, we are adjusting and improving, keeping readability high while also keeping the theming strong.
Story Part: Because the purpose of this project is data collecting, our client prefers the overall experience to be 20-30 minutes, In terms of story design in the early days of story design, we were inspired by Professor Jessica Hammer and decided to set the story background as a competition among colleagues in the company. In the following screenwriting, our producer repeatedly works with our project instructor to iterate our story, ultimately with our ETC story faculty’s help with the client making strong final adjustments and approving and locking this down.
What could have been better
We wish we could have more time to help our client, because due to the impact of the COVID-19 virus, Our client has been affected in the development progress and testing of APIs with our code (regarding the task questions and AI system and all the number data in the project) and art assets.Due to the timing of the funding approval for the client’s proposal for this work, and the remote COVID-19 situation, we were never able to integrate final story art, the API controlling tasks, and other elements authored by the client to make this complete. We are left hoping that our code will be documented well enough to allow the client to do such final integration.
Lessons learned and conclusion
As a producer, this is my first time as a producer of the ETC project. During the course of this semester, I learned the process of growing from 0 to 1 as a qualified producer, also I learned how to communicate with client, faculty and staff in a professional way, preside over the playtest process, how to listen to faculty and staff suggestions and make timely feedback. As a team, this is also the first ETC project of all four of our team, during this project, we completed the transformation from student to professional team. Our team deeply understood the operation process of a project development team, how to work with client and how to best help the client to achieve their goal, how to reconcile and arrange our workload.
In short, through this semester, our team of four has gained very valuable project development experience. We learned how to become a more successful ETCer. In terms of projects, our client decided to take over our project after this semester to complete the follow-up API and Art development. To this end, we have prepared detailed code documents for our client and cleared our code base to avoid redundancy and improve readability and maintainability.