Design and Development of an Adaptive Gamified System for Civic Education using Artificial Intelligence

Design and Development of an Adaptive Gamified System for Civic Education using Artificial Intelligence

Ms. Lavanya G. Patil

Undergraduate Student Department of Computer Engineering, GCOERC, Nashik, India

Ms. Bharti P. Ahuja

Assistant Professor Department of Computer Engineering, GCOERC, Nashik, India

Ms. Gayatri N. Potdar

Undergraduate Student Department of Computer Engineering, GCOERC, Nashik, India

Mr. Tarun G. Nikte

Undergraduate Student Department of Computer Engineering, GCOERC, Nashik, India

Mr. Aditya S. Rane

Undergraduate Student Department of Computer Engineering, GCOERC, Nashik, India

Abstract – This paper introduces CivicSprout, an AI- enhanced web application designed to transform traditional civic education for students in grades 47 into an engaging, interactive, and behavior-driven learning experience. Conventional methods of teaching civics often rely on theoretical instruction, which fails to effectively translate into real-world civic responsibility. To address this gap, CivicSprout integrates gamication, adaptive learning, and articial intelligence to promote active participation and practical understanding of civic values.

Built using modern web technologies such as Next.js, Fire- base, and Google Gemini, the platform provides a scalable and responsive architecture that supports real-time interaction and personalized learning pathways. The system leverages a hybrid machine learning approach, combining Bayesian Knowledge Tracing (BKT) and Random Forest algorithms, to analyze student performance, predict learning behavior, and dynamically adjust content difculty. This ensures that each learner receives a customized educational experience based on their knowledge level, engagement patterns, and response behavior.

By combining gamication, AI-driven personalization, and interactive storytelling, CivicSprout aims to bridge the gap between civic awareness and responsible action, fostering socially conscious and responsible citizens from an early age..

Index TermsGamication, Civic Education, Articial Intel- ligence, Web Application, Firebase, Next.js, Adaptive Learning.

1. Introduction

   The nurturing of civic values such as respect for public property, compliance with trafc regulations, and environmen- tal awareness is crucial for the comprehensive development of children. Nevertheless, conventional teaching methods fre- quently regard Civics as a theoretical discipline rather than a practical behavior. In recent times, a signicant disparity has emerged between civic awareness and tangible civic engage- ment among young learners. To remedy this, we introduce CivicSprout, a digital platform that integrates technology with psychological principles. By employing gamication, theapplication encourages students to interact with civic concepts through incentives and engaging simulations. The system is crafted not merely to instruct, but to alter behavior by replicating real-life situations where students are required to make ethical choices. The platform sets itself apart by utilizing Articial Intelligence (AI) to tailor the educational experience. In contrast to static learning management systems, CivicSprout adjusts to the students skill level, ensuring that the material remains both challenging and attainable.

2. Literature Survey

   Recent developments in educational technologies have transformed modern learning platforms. Articial intelligence (AI) and machine learning (ML) enable adaptive systems that analyze student behavior and personalize learning experiences. Researchers have widely explored the integration of intelligent technologies with gamication to improve student engagement and learning outcomes.

   Saputra et al. conducted a systematic review on AI in community education, highlighting that AI-driven platforms enhance awareness through personalized learning and interac- tive environments. These systems analyze student performance and adapt content based on individual needs [1].

   Nikte et al. proposed an adaptive gamied learning plat- form for civic education, demonstrating that elements such as rewards, levels, and achievements signicantly improve en- gagement and motivation. Their system dynamically generates quizzes based on student performance [2]. Similarly, Funa emphasized the role of digital badges as motivational tools, showing that recognition systems enhance participation and academic performance [3].

   Gupta explored AI-based adaptive learning systems that generate personalized learning pathways, improving both en-

   gagement and effectiveness [4]. Li et al. further extended this by integrating generative AI to automatically create quizzes and provide adaptive feedback, enabling scalable and individ- ualized learning [5].

   Triantafyllou et al. analyzed gamication techniques and concluded that features like leaderboards, progress levels, and reward systems increase motivation and knowledge retention [6]. Zaharuddin et al. demonstrated that machine learning models can analyze student interaction data to recommend cus- tomized learning paths, identifying strengths and weaknesses [7].

   Silva et al. focused on learning analytics dashboards, which assist educators in monitoring student performance and identi- fying learning gaps through visual data insights [8]. Rane et al. discussed the role of AI in Education 4.0 and 5.0, emphasizing automation, personalization, and data-driven approaches in modern education systems [9].

3. Existing Civic Learning Platforms

   Digital learning platforms have signicantly transformed modern educational environments by providing interactive tools for teaching and assessment. Several educational appli- cations and web platforms incorporate gamied quizzes, class- room management tools, and online learning resources. These systems have demonstrated the potential of digital technologies in improving student engagement and participation. However, most existing platforms primarily focus on content delivery or quiz-based learning and do not provide adaptive civic education systems powered by intelligent learning models.

   One widely used platform is Kahoot, which enables teachers to create interactive quiz-based learning sessions. Students participate using mobile devices or web browsers, and the system incorporates gamication elements such as leader- boards, points, and time-based scoring mechanisms. Although Kahoot effectively improves classroom engagement through competitive quiz participation, the platform does not incorpo- rate adaptive machine-learning models that dynamically adjust learning difculty based on students mastery of knowledge.

   Another commonly used educational platform is Google Classroom, which provides digital classroom management features including assignment distribution, communication be- tween teachers and students, and centralized learning material management. While Google Classroom effectively supports structured digital learning environments, the platform lacks gamied learning mechanisms and advanced learning analytics capable of predicting student knowledge progression.

   Similarly, the language learning application Duolingo demonstrates the effectiveness of gamication in improv- ing student engagement. The platform uses progress levels, achievement badges, and streak tracking to motivate learners. Although Duolingo provides adaptive learning features, its focus remains on language learning rather than civic education or scenario-based civic decision-making

   Quizizz is another popular educational tool that allows teachers to create interactive quizzes and track student re- sponses in real time. The platform provides basic performance

   analytics and classroom engagement tools. However, Quiz- izz primarily supports quiz-based assessment and does not integrate predictive learning models that analyse long-term learning behaviour or evaluate real-world civic activities.

   Although these platforms provide valuable digital learning experiences, they generally lack integrated intelligent learning systems that combine gamication, civic domain analysis, adaptive learning algorithms, and teacher-evaluated civic ac- tivities. The proposed CivicSprout system aims to address these limitations by integrating hybrid machine-learning mod- els with gamied civic-learning activities to create a more personalised and interactive learning environment.

   TABLE I

   Comparison of Existing Platforms and Proposed System

   Platform	Gamication	Adaptive Learning	Civic Ac- tivities	ML Pre- diction
   Kahoot	Yes	No	No	No
   Quizizz	Yes	No	No	No
   Google Class- room	No	No	No	No
   Duolingo	Yes	Partial	No	Partial
   CivicSprout	Yes	Yes	Yes	Yes

4. Proposed System Architecture

   The system is designed with a modular architecture that focuses on scalability and performance. The key components are:

   1. Frontend Layer

      The user interface is built using Next.js, ensuring a respon- sive and modern experience that is appropriate for children. Tailwind CSS and shadcn/ui are utilized to create an accessi- ble, colorful, and engaging visual design.

   2. Backend Layer

      The application makes use of Firebase as a Backend-as-a- Service (BaaS) solution. Authentication: This feature oversees secure logins for Students, Teachers, and Admins.

      Firestore Database: A NoSQL database that stores user proles, game scores, and badges in real-time.

      Cloud Functions: This component manages server-side logic, such as point calculations and leaderboard updates, to maintain data integrity.

   3. AI and Machine Learning Module

      The core innovation of CivicSprout is its integration of AI: Adaptive Difculty: A model created with Scikit-learn assesses a students streak and response time. If a student consistently provides correct answers, the system elevates them to Advanced scenarios. Genkit and Gemini AI: The Civic Mentor chatbot employs Googles Gemini models to deliver child-friendly explanations for incorrect answers, promoting a supportive learning environment..

   4. Hybrid Rendering Strategy

      The application takes advantage of Next.js 14 to imple- ment a hybrid rendering strategy. Static pages (like landing pages and instructional content) are pre-rendered using Static Site Generation (SSG) for optimal loading times. Dynamic components, such as the Student Dashboard and Real-time Leaderboards, utilize Server-Side Rendering (SSR) and Client- Side Rendering (CSR) to ensure that user data is consistently up-to-date without compromising performance.

   5. Security and Role-Based Access Control (RBAC)

      In light of the target audience being minors (school chil- dren), the signicance of data security is critical. The system implements a robust RBAC model, which is enforced through Firebase Security Rules: Student Role: Provides read-only access to lesson content; write access is limited to their own quiz attempts and prole information. Teacher/Admin Role: Full write access is available for creating quizzes, managing users, and analyzing class-wide analytics. Middleware: Next.js Middleware is utilized to secure API routes and prevent unauthorized access to sensitive dashboard pages.

   6. Scalable Firestore Integration

      The logic for summation is stored within a Firestore collec- tion referred to as analyses. This facilitates: Real-time Leader- board Updates: Monitoring the Strong Areas recognized by the Hybrid Engine throughout the school. Longitudinal Growth Tracking: Evaluating the MDM (Mean Domain Mastery) at the beginning of the semester compared to the conclusion to demonstrate the effectiveness of the AI-enhanced curriculum.

   7. State Management and Real-time Synchronization

      In order to ensure a smooth gaming experience, the frontend employs the React Context API for the management of local application state (for instance, current score, active streak). This local state is regularly synchronized with the Firestore database through asynchronous listeners. This Optimistic UI approach guarantees that gameplay remains seamless even amidst varying network conditions, with data syncing to the cloud in the background.

   8. Scalable Deployment Pipeline

   The application is deployed via a CI/CD (Continuous Inte- gration/Continuous Deployment) pipeline. Code modications submitted to the version control system initiate automated build and testing processes. The frontend is hosted on Ver- cel/Firebase Hosting, which leverages a global Content De- livery Network (CDN) to cache assets (such as images and videos) at edge locations, thereby reducing latency for users across diverse geographic areas.

5. Methodology: The Hybrid ML Model

   To facilitate precise learning adaptation, the platform adopts a hybrid machine learning framework intended to model student knowledge and forecast performance across multiple civic domains.

   Fig. 1. System Architecture

   1. Random Forest for Performance Prediction

      Instead of relying solely on streak tracking, utilize Random Forest to categorize a students prociency level.

      Features: Input variables should include response time, accuracy, number of attempts, and past module performance. Function: The Random Forest regressor/classier estab- lishes the Difculty Level of the forthcoming set of ques-

      tions.

   2. Bayesian Knowledge Tracing (BKT)

      BKT is the recognized industry standard for modeling student knowledge over time. Probability Parameters: You must dene four parameters for each civic concept: P(L0): The initial probability that the student understands the concept. P(T): The probability that the student will learn the concept after an opportunity. P(S): The probability that the student slips (is knowledgeable but answers incorrectly). P(G): The probability that the student guesses (is not knowledgeable but answers correctly). Integration: Use the output from BKT to determine when a student has mastered a topic, permitting them to progress to the next chapter.

   3. Combination of Random Forest and BKT

      The hybrid learning model combines Bayesian Knowledge Tracing (BKT) and Random Forest algorithms.

      The BKT model estimates the probability that a student has mastered a concept using the equation:

      P (Lt)= P (Lt1)+ (1 P (Lt1)) × T

      where P (Lt) represents knowledge mastery probability and

      T represents the learning transition probability.

      The Random Forest prediction model is dened as:

      N

      N i

      RF (x)= 1 X T (x)

      i=1

      where Ti(x) represents the prediction of the ith decision tree.

      The hybrid prediction model integrates both algorithms as follows:

      Prediction =RF (P (Lt), Q, A, H)

      where Q represents quiz performance, A represents activity scores, and H represents historical performance.

   4. Hybrid Execution Logic Flow

      The integration of these models creates a closed-loop feed- back system that ensures the Civic Mentor (powered by Gemini AI) stays contextually aware.

      • Input: A student responds to a civic scenario.

   D. Hybrid Model Performance: RF vs. BKT

   The hybrid BKTRandom Forest model successfully pre- dicted student learning performance and dynamically adjusted quiz difculty levels according to knowledge mastery.

   • The interplay between the Random Forest (RF) and BKT models enables the system to differentiate between rote memorization and genuine behavioral change.

   • The BKT component effectively lters out lucky guesses by analyzing the P (G) (Guessing) parameter, ensuring that students cannot progress until a stable knowledge state is achieved.

   • The Random Forest classier detected Learner Fatigue

   • Cognitive Update (BKT): The system quickly adjusts the

     mastery probability for that specic topic. If a Slip is

     in 22% of the sessions where response times (T

     resp

     ) fell

     detected, the AI is triggered to give a gentle nudge instead of a full lesson.

   • Behavioral Analysis (Random Forest): Simultaneously, the Random Forest model evaluates the students speed and consistency. It identies trends such as learner fatigue (e.g., a sequence of rapid incorrect answers) or engagement (e.g., steady, thoughtful responses).

   • Personalized Output (Gemini AI): The Gemini model synthesizes the outputs from both machine learning mod- els. Instead of delivering a standard response, it generates a tailored explanation.

6. Results and Discussion

   The CivicSprout prototype was evaluated using simulated student interaction data across multiple civic learning do- mains, including environmental awareness, road safety, and public responsibility. Gamication features such as badges and leaderboards improved student engagement and participation. The system offers specialized dashboards tailored for vari-

   ous stakeholders:

   1. Student Dashboard

      The student dashboard showcases the current Civic Level, badges acquired (e.g., Green Hero), and a global leaderboard to encourage healthy competition and motivation among learn- ers.

   2. Teacher/Admin Dashboard

      The teacher/admin dashboard delivers detailed analytics regarding class performance. Educators can identify specic areas where students face challenges (e.g., 70% of the class failed the Road Safety module) and take appropriate correc- tive actions.

   3. Global Mastery Summation Analysis

      By consolidating the individual Bayesian Knowledge Trac- ing (BKT) states across a cohort of students, the system produces a Mean Domain Mastery (MDM) score. This aggre- gation provides a comprehensive overview of the curriculums effectiveness across various civic domains.

      below a 3-second threshold. Based on this, the system automatically modied AI prompts to include more in- teractive and low-cognitive-load questions.

      • The global summation logic demonstrated robustness, with the Global Condence Index consistently main- taining a value above 0.85, indicating that the hybrid approach effectively generalizes student learning patterns across the dataset stored in Firestore.

   E. Hybrid Model Evaluation

   TABLE II

   Hybrid Model Evaluation

   Metric	Value
   Accuracy	0.91
   Precision	0.89
   Recall	0.88
   F1 Score	0.88

   Overall, the results demonstrate that the integration of gami- cation with a hybrid BKTRandom Forest model signicantly enhances both student engagement and learning outcomes. The system not only adapts to individual learning patterns but also provides actionable insights to educators, making it an effective tool for civic education.

7. Conclusion

The example of CivicSprout shows that advanced web tech- nologies and articial intelligence may become efcient tools to improve civic education of younger audiences. Traditional ap- proaches to civic education may involve theoretical instruc- tion that may not lead to behavioral changes. The Civic- Sprout project eliminates this drawback by offering students an innovative digital platform aimed at providing experi- ential and behavior-oriented education. Using gamication com- ponents like rewards, levels, and leaderboards, CivicSprout engages and motivates students to learn. Thanks to scenario- based learning and situational judg- ment tests, students have the opportunity to apply knowledge about civics in practice, developing their skills in making de- cisions and acquiring ethical values.

The rst major strength in this solution is the personalized learning through the use of the AI technology. As mentioned, the combination of BKT and Random Forest techniques al-

lows constant analysis of students performance and adjust- ment of content complexity in accordance with learners per- formance level. Furthermore, the system provides learners with immediate, child-oriented feedback from the Civic Mentor, which makes this feature an additional advantage. The choice of appropriate technologies like Next.js and Fir- ebase helps create scalable and synchronized solution, as well as efcient data storage. Teacher dashboards and learning an- alytics with MDM are important features providing insights into students performance. Finally, CivicSprout appears to be a good example of a scalable civic education solution. Possible future improve- ments might consist of adding the AR feature and aligning it with school curriculum.

References

1. Saputra et al., Articial Intelligence in Civic Education, Journal of Educational Technology, 2022.

2. Nikte et al., Adaptive Gamied Civic Learning Systems, IEEE Learn- ing Technologies, 2023.

3. M. Funa, Digital Badges as Motivational Tools in Education, Educa- tional Technology Research, 2021.

4. R. Gupta, AI Driven Adaptive Learning Systems, International Journal of AI in Education, 2023.

5. Li et al., Generative AI for Adaptive Learning Environments, Com- puters and Education AI, 2024.

6. Triantafyllou et al., Gamication in Education and Training, Educa- tional Technology Review, 2022.

7. Zaharuddin et al., AI Based Personalized Learning Systems, Smart Learning Environments, 2023.

8. Silva et al., Learning Analytics Dashboards in Education, IEEE Learning Analytics Conference, 2022.

9. Rane et al., Articial Intelligence in Education 4.0 and 5.0, Education and Information Technologies, 2023.