Hello everyone. I am Ozaki, a first-year Master’s student. In this article, I would like to introduce the content and my thoughts regarding the paper I read in our most recent English Literature Seminar.

• Paper Title: CareerSim: Gamification Design Leveraging LLMs For Career Development Reflection

• Journal and Pages: Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (pp. 1-7)

• Year of Publication: 2024

• Authors: Wantong Du, Zhiying Zhu, Xinhui Xu, Haoyuan Che, Shi Chen

1. Background and Objectives

In recent unstable social conditions, self-construction, decision-making, and long-term perspectives are becoming important for career development. However, effectively acquiring these skills is difficult, and a lack of information and understanding poses a challenge. In this context, “gamification” is drawing attention as a creative approach in the field of career education; however, concerns have also been raised that it might oversimplify reality or neglect reflection. Regarding these issues, by building a comprehensive database and utilizing Large Language Models (LLMs) in design, it may be possible to generate more personalized and deep narratives and produce immersive user experiences, thereby promoting meaningful reflection.

In this study, the design, development, and preliminary investigation of a career education simulation RPG titled “CareerSim,” currently under development, were conducted. The objective is to “strengthen the relevance between game events and the real world as well as reflection through the utilization of LLMs.”

2. Game Design of CareerSim

In the design process, the researchers referred to the “DMC Pyramid Model (Wiklund and Wakerius, 2016),” classifying the design elements into three perspectives: Dynamics, Mechanics, and Components.

(1) Dynamics Perspective and Game Genre CareerSim is designed as an educational RPG. Educational RPGs have the characteristic of providing player progress as a record for reflection. Players experience a 30-year career through a character, exploring an unpredictable future and their own professional identity. During play, players can construct a virtual character and confirm their career development.

(2) Mechanics Perspective CareerSim is designed based on the “Progression Mechanism (Ljungkvist and Mozelius, 2012).” Player career formation is divided into stages, with themes, events, and tasks set for each stage. Players must explore career possibilities and make decisions at each stage. Super’s Career Development Theory provides the theoretical support for this design. Decision-making within the game plays a role in prompting the player’s real-world reflection. Additionally, a “Chance Mechanism” was designed based on career theories, taking into account randomly occurring social, economic, and cultural external factors.

(3) Components Perspective To visualize the results of decision-making, characteristic settings such as “Realistic, Investigative, and Social” are incorporated. These characteristics are based on Holland’s Theory of Vocational Personalities and are defined by the player at the start of the game. During game progression, these characteristics change dynamically based on the player’s decisions and also influence the success rate of those decisions. Furthermore, by adopting a low-pixel interface, the design ensures focus on the story and maintains convenience for educational gamification.

3. Game Concept

CareerSim is playable on a PC, where the player assumes the role of a virtual character and sets various characteristics in the initial stage. By experiencing events related to career development and forming a virtual career, they gain opportunities to reflect on self-construction and career decision-making.

Regarding the game’s worldview, it begins with the player graduating from University X as an Industrial Design major. As they form their career through events such as further education or employment, careful choices are required. Players develop their characteristics and roles during the decision-making process, retire at age 55, and receive a report regarding their career. The game progresses through career formation events and is largely divided into four phases:

(1) Character Setting: At the start of the game, the character’s characteristics and the city of residence are set. By combining characteristics and vocational personality theory, the initial optimal career position is matched. Characteristics are dynamically adjusted through events and player choices. (2) Career Development Events: After starting the game, players experience career development events generated by the system. These events are influenced by the player’s career development stage and individual characteristics. (3) Decision-Making and Impact: Players consider options within the game and make decisions. This changes the character’s characteristics, and the age is updated within the range of the next stage. These changes influence subsequent events. (4) Post-Completion Report: Upon completion of the game, a comprehensive career development report is created. It is designed to allow players to look back on the events that occurred and prompt reflective thinking.

4. Functional Implementation

The main functions of CareerSim are divided into the following four areas:

(1) Input: There are two elements. The first is the characteristic input at the start of the game, which functions as a trigger for the initial event generation. The second is the player’s decisions during game progression, based on which characteristics are dynamically modified. (2) Database: Composed of actual career events of graduates from a certain university. These data are organized based on Super’s Career Development Theory and classified based on Holland’s Theory of Vocational Personalities. (3) Career Event Generation: Utilizing Langchain and a Large Language Model (GPT-4), individual and authentic events are generated. Events within the database are divided into vectors and searched using cosine similarity comparison to find career events that match the player’s input. Players are proposed multiple options, including random events generated by AI, and make decisions from among them. (4) Interaction with Player: Real-time interaction is made possible through Dynamic Feedback and interface design using Unity. Based on the player’s characteristics and decisions, the next career events and options are generated. Additionally, a memory module is used to record the player’s interactions, which then influences subsequent events.

5. Preliminary Investigation Results

The preliminary investigation was conducted with 12 university students aged 18 to 24. After they actually experienced CareerSim, semi-structured interviews were conducted. From the survey results, it was found that CareerSim can produce an immersive experience including diverse decision-making. Users engaged with initial characteristic settings and decision-making in the game while integrating them with real-world situations, showing that the system design prompted personalized career exploration (for example, evaluations such as “it’s realistic, I can objectively understand future scenarios, and I can gain early experience and preparation”).

On the other hand, comments like “When I was a new graduate, I had a clear plan for career choice, but at age 40, I no longer knew what factors to consider when making decisions” indicated that the immersion and relevance of the game might decrease with age.

Additionally, it was confirmed that changes in characteristics led to player reflection. Specifically, by seeing decisions that did not go as expected or decreases in characteristic values due to experiences of “cruelty” or “failure,” users gained opportunities to look back on their choices (for example, evaluations such as “I don’t usually think about career pitfalls, but facing failure and setbacks in the game made it easier to reflect on my choices”). This shows that adversity within the game has the potential to enhance real-world reflection and resilience.

In conclusion of the investigation, the effectiveness of a system and gamification design that induces career reflection was demonstrated. However, a challenge also emerged: in scenarios predicting the distant future, there are limits to the depth of the background setting, making it difficult to provide individual and authentic answers. Future development suggests the need to introduce richer themes, such as family factors, into the career development stages.

6. Conclusion and Future Challenges

Through development and preliminary investigation, it was confirmed that utilizing LLMs provides appropriate feedback for career exploration and has the effect of increasing motivation for reflection and other activities.

Future improvements include integrating real-time Internet databases to create the latest career scenarios and refining LLMs based on career development knowledge to create more realistic causal relationships. Future plans include measuring users’ experience and understanding of self-construction using questionnaires and interviews. Additionally, through repeated version upgrades, the study aims to pursue the potential of gamification across entire learning domains and contribute to LLM-based gamification theory.

My Thoughts

I read this paper because I was interested in the theme of whether gamification can be utilized to lead to personal career reflection. From a game design perspective, I found it very helpful that the use of LLMs and data from 50 actual graduates enabled the construction of stories on a large, realistic, and individual scale. Furthermore, I believe that by being able to easily verify what kind of result would have occurred had a different choice been made within the system, it might be possible to perceive one’s own decision-making objectively through comparison, leading to deep reflection. On the other hand, the game genre this time was an RPG, which assumes individual learning. Since personal career reflection requires understanding of others and society as well as self-understanding, I became curious about the advantages in cases where collaborative or group learning styles are taken as the game system. In the future, regarding the field of career-related game design, I would like to conceptualize designs while considering more diverse genres and utilizing technologies according to the purpose and target.