Bridging Pedagogy and Play: Introducing a Language Mapping Interface for Human-AI Co-Creation in Educational Game Design Conference: Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems; April 13–17, 2026; Barcelona, Spain Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems (CHI EA ’26), April 13–17, 2026, Barcelona, Spain DOI: 10.1145/3772363.3798862 ISBN: 979-8-4007-2281-3/2026/04 CCS: Human-centered computing Human computer interaction (HCI) CCS: Applied computing Interactive learning environments CCS: Applied computing Computer games

Daijin Yang, Erica Kleinman, Casper Harteveld · 2026

Read Bridging Pedagogy and Play: Introducing a Language Mapping Interface for Human-AI Co-Creation in Educational Game Design Conference: Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems; April 13–17, 2026; Barcelona, Spain Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems (CHI EA ’26), April 13–17, 2026, Barcelona, Spain DOI: 10.1145/3772363.3798862 ISBN: 979-8-4007-2281-3/2026/04 CCS: Human-centered computing Human computer interaction (HCI) CCS: Applied computing Interactive learning environments CCS: Applied computing Computer games by Daijin Yang, Erica Kleinman, Casper Harteveld (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-03644.txt.

Introduction

Daijin Yang, Erica Kleinman, and Casper Harteveld give teachers a shared language for designing games with artificial intelligence. Educational games build thinking skills and motivation, but instructors struggle to design games that hit specific learning goals. Existing tools hide the hard parts behind opaque machine suggestions. Their 2026 paper presents a web tool where language itself is the interface for co-creation.

Inside, a teacher and a language model assistant jointly develop a structured language mapping pedagogy onto gameplay through four linked components. Every pedagogical intention stays explicit and editable on screen. The authors argue this lowers design barriers for non-experts and keeps human agency over the critical decisions. It also lets designers check alignment between teaching goals and game mechanics during and after creation, instead of discovering mismatches in the classroom.

Catalog record

CREATORDaijin Yang, Erica Kleinman, Casper Harteveld
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE19366
LICENSEhttp://creativecommons.org/licenses/by-sa/4.0/
RIGHTSCC BY SA
SOURCEarXiv

Source: https://arxiv.org/abs/2603.03644

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Bridging Pedagogy and Play: Introducing a Language Mapping Interface for Human-AI Co-Creation in Educational Game Design Conference: Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems; April 13–17, 2026; Barcelona, Spain Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems (CHI EA ’26), April 13–17, 2026, Barcelona, Spain DOI: 10.1145/3772363.3798862 ISBN: 979-8-4007-2281-3/2026/04 CCS: Human-centered computing Human computer interaction (HCI) CCS: Applied computing Interactive learning environments CCS: Applied computing Computer games Daijin Yang, Erica Kleinman, Casper Harteveld Abstract . Educational games can foster critical thinking, problem-solving, and motivation, yet instructors often find it difficult to design games that reliably achieve specific learning outcomes. Existing authoring environments reduce the need for programming expertise, but they do not eliminate the underlying challenges of educational game design, and they can leave non-expert designers reliant on opaque suggestions from AI systems. We designed a controlled natural language framework-based web tool that positions language as the primary interface for LLM-assisted educational game design. In the tool, users and an LLM assistant collaboratively develop a structured language that maps pedagogy to gameplay through four linked components. We argue that, by making pedagogical intent explicit and editable in the interface, the tool has the potential to lower design barriers for non-expert designers, preserves human agency in critical decisions, and enables alignment and reflections between pedagogy and gameplay during and after co-creation. 1. Introduction Educational games can enrich education by supporting critical thinking, problem solving, and collaboration in interactive learning environments ( 4 ; 35 ; 22 ; 7 ; 8 ) . Empirical studies further show that well-designed educational games can increase learners’ motivation and retention of disciplinary content ( 4 ; 46 ) . However, educational games often fall short when game experiences do not align with course content, assessment practices, or classroom routines, which can reduce pedagogical value and introduce conceptual confusion ( 18 ; 19 ; 25 ) . These misalignments contribute to instructors’ continued hesitancy to adopt educational games systematically ( 29 ; 9 ) . Although instructors understand learning objectives and student needs, many lack training in game and learning experience design, and they face persistent challenges in configuring mechanics, goals, narrative, and user experience to serve specific learning outcomes without sustained scaffolding ( 5 ; 2 ; 14 ; 20 ) . Low-code authoring environments can lower programming barriers, but they do not resolve these deeper design and game-literacy demands ( 39 ; 33 ; 13 ; 14 ; 20 ) . Large Language Models (LLM) offer a possible response by assisting instructors across design stages, from early ideation to prototype implementation, and by enabling mixed-initiative co-creative workflows ( 9 ; 44 ; 23 ; 10 ; 45 ) . By generating candidate content, code, and prototypes from natural language or structured design documents, these tools can reduce manual programming and help instructors translate pedagogical intentions into playable experiences ( 44 ) . However, naive use of LLM assistants introduces risks for instructors. An instructor may receive a seemingly complete game with limited insight into how it supports stated learning objectives, which content elements are fabricated, or how design choices may exacerbate challenges for students who already struggle with gameplay or disciplinary understanding ( 25 ; 6 ; 18 ) . Prior work on LLM-supported design cautions that opaque automation can diminish human agency by forcing instructors to reverse engineer under-specified suggestions or by relegating them to content approvers rather than designers of educational intent ( 23 ; 45 ) . Consequently, AI-supported educational game design requires communication that foregrounds transparency and fine-grained control, enabling instructors to inspect, critique, and iteratively adjust AI proposals while maintaining alignment with course goals and diverse learner needs ( 2 ; 5 ; 27 ) . We address these needs by designing a web interface that operationalizes a language-based framework as the central collaboration interface for LLM-assisted educational game design. Instructors and the system co-author a short, structured description of a learning activity, and the interface generates a corresponding game-level description using the same sentence structure. By making this shared representation editable and inspectable, the tool links pedagogical intent to concrete game decisions about rules, content, difficulty, and context, while providing stable handles for constraint setting, explanation, and revision. In doing so, the tool shows the potential to instantiate a controlled language-mediated communication protocol for human–AI co-creation that supports goal negotiation, traceability, and responsibility in generative design workflows. 2. Related Work Prior work establishes both the promise and the practical difficulty of educational games, and it motivates approaches that embed Large Language Models (LLM) within transparent, instructor-centered co-creative processes. Although educational games can support motivation and learning through interactive practice and feedback ( 4 ; 41 ; 46 ) , instructors often hesitate to adopt them when game experiences misalign with course content, assessment practices, and classroom routines, or when suitable resources and institutional support are limited ( 25 ; 18 ; 19 ; 26 ; 9 ) . Even when instructors wish to create or adapt games, they frequently lack preparation in game and learning experience design, and they must fit design work into constrained instructional time ( 5 ; 27 ; 2 ) . Recent LLM-based game design tools appear to lower technical barriers by generating ideas, rules, assets, and prototypes from prompts or structured descriptions ( 17 ; 42 ; 16 ; 43 ; 12 ) . However, studies of LLM-supported generation also highlight persistent concerns about limited controllability and transparency, which can shift effort toward verification and repair to address incoherence, inaccuracies, or weak pedagogical alignment ( 43 ; 16 ; 42 ; 40 ) . Educators similarly report concerns about hallucinations, misalignment with objectives, and increased checking workload, which can undermine pedagogical control even when productivity improves ( 15 ; 32 ; 30 ) . Human–AI co-creation frameworks help structure collaboration through roles, phases, and initiative, but they often provide limited guidance on how instructors should communicate intent, constraints, and rationales to support inspection and revision during design ( 24 ; 1 ; 28 ; 37 ; 36 ) . As one of the possible solutions, controlled natural languages demonstrate how constrained, readable text can reduce ambiguity while remaining machine parsable, enabling domain experts to author, inspect, and revise specifications with computational support ( 21 ; 11 ) . Analogous language-oriented representations in game design, including the Video Game Description Language and the Learning Mechanics–Game Mechanics model, further show a potential way how structured descriptions can expose design choices and connect mechanics to learning processes in ways that support reflection and iteration ( 38 ; 3 ; 34 ) . Taken together, the literature motivates instructor-centered co-creative tools that use language-mediated representations to make pedagogical intent explicit, inspectable, and revisable while leveraging LLM generation. 3. Introducing the educational game design language mapping framework Based on the literature, we propose a language mapping based framework. The framework represents educational intent as a single controlled sentence that is readable to humans and parseable by the system, and acts as the co-creation interface: “Players (Students) [ Adverbs ] [ Verbs ] [ Nouns ] in a [ Adjectives ] environment.” This template is the basic unit

Read the complete text