Getting to know machine learning through game – Haram Kwon, Joonhyung Bae, Karam Eum, and Seolhee Lee at KAIST Graduate School of Culture Technology
NC has been cooperating with KAIST Graduate School of Culture Technology since 2020 to foster future R&D talents and develop game contents research. Best work in 2021 of NC-KAIST game design project conducted as part of industry-academia collaboration has won in the ‘Transformative and Transgressive Play’ category at Student Game Competition of 2022 ACM Conference on Human Factors in Computing Systems(CHI) held in April to May(Link).
In this interview, NC Blog met the student team(Haram Kwon, Joonhyung Bae, Karam Eum, and Seolhee Lee) of KAIST Graduate School of Culture Technology which developed ‘Classy Trash Monster(CTM)’, the winner of the competition, and listened to the game development process in detail. In addition, the class reviews by Professor Joohan Nam of KAIST Graduate School of Culture Technology who lead the game design project class and by Donggyo Lee, Director of NC Game Design Lab in charge of practical lecture and feedback will be released.
From left Haram Kwon, Seolhee Lee, Karam Eum, Joonhyung Bae
Viewing social issues through game narrative
Please introduce Classy Trash Monster which is the winner of this Student Game Competition.
CTM Development Team) Classy Trash Monster is a game designed to help non-majors with no background knowledge on machine learning naturally learn it. Machine learning is often used in everyday life, and there are more and more non-major students who want to learn it. Therefore, we wanted to make a game giving people without expert knowledge confidence to overcome the barriers and prejudices they encounter while studying the machine learning and to continue studying.
What does the name of Classy Trash Monster game mean?
(CTM Development Team) We named the game Classy Trash Monster(CTM) so that people may intuitively know that it is a game classifying trash monsters by training a model. Trash monster symbolizes an environment issue not solved by the society. If a player succeeds in classification, a recycling monster is born. The name ‘Classy Trash Monster’ embeds hope that a player would experience the process of solving environment issues through game narratives and machine learning model.
I wonder the roles each of you played.
(Joonhyung Bae) While developing CTM, I focused on two experiences that users can feel. First of all, I wanted to convey experiences that a player can learn about the relationship between machine learning and data while playing the game and feel how a machine learning classification model can contribute to society through the narrative of CTM. So, I thought a lot when designing the artwork.
(Karam Eum) I was responsible for game level design and play test. When designing and conducting the play test, I aimed to find things to be improved in the game based on the actual experience of player. I verified how much the player’s experience intended in the development stage is the same as what a player actually experience while playing the game. I aimed to prove how the player’s experience intended in the development stage is consistent with player’s actual experience, and improve inconsistencies.
(Seolhee Lee) I was in charge of UI part in the CTM. I applied the outcomes of artwork to the game and tested what control feel the designed UI gives when it is actually operated. I focused on composing UI/UX that makes a player intuitively understand and be immersed in.
(Haram Kwon) As I was responsible for main development, I realized model training, game elements, and event system, etc. I strived to make casual play possible without damage to underlying knowledge while integrating the elements of machine learning into game elements through an idea such as data collector that can capture trash monsters with simple number keys and simultaneously label them.
Your game won on the ‘social impact of games’ category of the ACM CHI competition. In what ways do you think CTM has social value?
(CTM DEVELOPMENT TEAM) The evaluation in the ‘social impact of games’ category of the ACM CHI focuses on how much the role of game in areas such as education, economy, and social culture has been explored by expanding the boundaries of game. With the rise of artificial intelligence technology, data literacy related to this is becoming increasingly important in our society. I think this game has received attention since it was played it in the educational aspect by utilizing the game’s own characteristics and strengths.
Thankfully, our game received comments on being aesthetically beautiful at the time of announcement at the CHI competition. I think they empathized with the play experience and message we were trying to convey by harmonizing game design, development, and artwork.
Balancing fun and learning
It is also the result of game design project class in collaboration between KAIST and NC Game Design Lab. I wonder what the class was like with practitioners in the field.
(Karam Eum) I am considering a career as a game researcher. The industry-academia collaboration class was an opportunity to expand my insights through the interaction with game developers in the field. The feedbacks I received during the interaction really helped set the direction of our research. The lectures prepared by Game Design Lab are particularly memorable. I heard a lot of practical advice on game development, such as how to incorporate the concept of fun into the output and how to prioritize in the development process.
(Haram Kwon) I thought I had played quite a few games and built up my own reference, but the point of view presented by the Game Design Lab practitioners has clear references and is systematic, and so I felt I was not good enough. I thought I should think more about how to design the game and give fun to players.
(Joonhyung Bae) I participated in the class since I wanted to develop impact game. It was meaningful to learn the knowledge that I can use while actually developing games. I was stimulated a lot by seeing how much NC Game Design Lab practitioners love games. I also enjoyed the process of team members working together with one goal.
What is the most memorable comment you received from the NC staff during the game development process?
(Seolhee Lee) The most memorable one is the feedback that fun element can conflict when genres are mixed. It was helpful to develop game elements and UI while thinking about the balance of interactions within the game.
(Haram Kwon) CTM was originally designed based on a genre close to construction management rather than defense. However, more additional elements had to appear to give the genre's fundamental fun, and consequently, it was far from education on machine learning, which was an ironic situation. And then, I was able to develop a game that every element was harmonized thanks to the change of genre to defense after receiving Game Design Lab’s comments 그러던 중 Game Design Lab.
(Joonhyung Bae) The feedback that I should consider the interaction between the space where the recycling machine exists on the left of the game screen and the control panel UI on the right was most memorable. It was a great help in UI/UX design to quickly catch changes in the development process and give feedback to us.
(Karam Eum) Although it is an educational game, we also wanted to have a fun element. When we were considering the balance between education and fun, Game Design Lab advised us to focus on 'design that can continuously motivate players'. In fact, the game design was constantly revised during development, and we were able to consistently design the player experience with the advice as a pivot. I think we were able to receive positive feedback from players thanks to this. The key to multiplayer game is 'fun imbalance'. We just focused on balancing, but after realizing that it’s fun when balance and imbalance are repeated, we could change the direction of development.
What kind of game or service do you want to make in the future?
(Haram Kwon) If the metaverse is the digital extension of the real world, I think a game is the creation of another reality in the digital world. I want to make games of various genres. But ultimately, I want to make a game like a small world where people can live an independent and unique life.
(Joonhyung Bae) Currently, I am researching methods to create motion from sound and visualize multi-modal data centering on sound data at the Music and Audio Computing Lab. I plan to study the relationship between formative languages in various art domains and share them through papers and artwork in the future.
(Karam Eum) I am researching self-actualization experiences and back-to-normal through games. I’d like to focus on the process of transferring emotions and thoughts experienced in games to everyday life, and share the research results that give answers to ‘What is the healing that only game play can provide?’.
(Seolhee Lee) I am researching AI automatic composition at the Music and Audio Computing Lab. My goal is to research AI composers who can show high-level artistry.
Completing the final puzzle to the game with practical experience
Donggyo Lee, Director of NC Game Design Lab
What is the Game Design Lab’s role at NC?
(Donggyo Lee) The Game Design Lab plays a role in researching and proposing design methods to discover design issues in advance in collaboration with NC's new projects. If there is a game of a new genre that we have not yet challenged, we analyze and systematize the essential design elements that each genre must have. Through the industry-academia research project with KAIST Graduate School of Culture Technology, it is also conducting a project to find game design factors that have achieved great results in the market and systematically identify their principles.
What kind of role did the Game Design Lab play in the game design project class?
(Donggyo Lee) Game Design Lab practitioners reviewed the results of the development process from game ideas generated after the lecture once a month and gave feedback. At the end of the project, it exchanged opinions and ideas in writing and communicated with the students so that they could refer to it for final polishing.
As Classy Trash Monster, which won this competition, was a game developed with a very challenging and novel idea, we had a lot to think about at the time of feedback. Our feedback was focused on how to interact with players and how to design if we were to gamify machine learning.
How was the feedback for the students done in detail?
(Donggyo Lee) In the first year of the game design project class, I was able to visit the school directly, make eye contact, give feedback and interact with each other. But last year, all lectures had to be conducted non-face-to-face due to COVID-19.
In the course of the class, I felt KAIST students already had professional-level programming skills. However, since it was the first time to make a game using this technology, I wanted to focus on the game a little more and let them know how to solve the various difficulties they may encounter in the development process through game design.
To put it simply, I suggested them to think with an approach of opening a buffet restaurant that can satisfy the tastes of many people rather than a well-planned course meal. While explaining the core game design principles, I tried to convey the key point that the processes to achieve various goals should be designed in advance to elicit various attempts from players.
What kind of game is the Classy Trash Monster, the winner of the competition remembered as?
(Donggyo Lee) As Classy Trash Monster was a game developed with a very novel and challenging idea, it was so interesting. At the same time, it was a project that made us think a lot about how to interact with players and design a game if machine learning was to be gamified. I remember studying machine learning without students’ knowing. (smile)
What contribution can this industry-academia collaboration class make in terms of supporting future technical talents?
(Donggyo Lee) I feel most grateful and proud when students who have taken the class express their ambitions to do an internship in order to gain experience in game design at NC or that they want to work in game making in the future. I think the greatest harvest of the industry-academia collaboration is to make game players know the joy of game itself and want to become a game maker at NC. In addition, meeting with students provides fresh ideas to me and other Game Design Lab practitioners. It is also meaningful in that it allows you to recall things that you would not normally think of.
Creativity of future generation is making emerging opportunities
What did you focus on most while teaching a game design project class with NC for the past two years?
(KAIST Joohan Nam) I guided the students’ game design in two major directions. The first one is the application of new interfaces and technologies that can overcome the limitations of existing gameplay methods. The second one is to deal with various topics such as education, health, safety, art, economy, and social meaning through games, beyond simple fun.
What aspects of the game design project class do you think have been helpful to the students?
(KAIST Joohan Nam) We have carried out special lectures by NC practitioners with long experience and feedback by student teams several times. I think it was very helpful for students to receive advice on the essential elements of game in the process of turning their initial ideas into a game in action. Of course, a wealth of advice on the details of game design, such as interaction, visuals, and sound, is also included.
What areas would you like to focus on to foster R&D talent at KAIST in the future?
(KAIST Joohan Nam) Although the role of education is important for nurturing future talent, partnerships with companies with successful experience in the industry like NC are also essential.
In addition to ongoing industry-academia research with NC, through industry-academia collaboration class such as the game design project, we plan to provide more opportunities so that students who are future generations may share their experiences and knowledge to realize ideas and creativity through. I think that the living experience and knowledge of the industry are transmitted to schools through classes, and innovative games made by students through this are also the positive result of industry-academia collaboration.