AIFE 2026 SINGAPORE: AI FOR EDUCATION SINGAPORE CONFERENCE 2026
PROGRAM FOR WEDNESDAY, NOVEMBER 18TH
Days:
previous day
all days

View: session overviewtalk overview

09:00-18:00 Hotdesking Room

Conference participants may make use of the facilities for quiet work.

Location: NIE7-TR714
09:00-10:00 Keynote 4: Professor Carolyn Penstein Rosé

New Learning Opportunities with AI Agent Facilitators of Collaborative Learning

Synopsis

For decades, Dr. Rosé has researched the use of AI technologies in service of human collaboration and human learning, and currently focuses on optimization of human-AI collaboration in order to speed the way to increased job opportunities and economic growth by creating actionable knowledge related to new roles for humans and AI agents instead of feared job displacement and unemployment. Her work has shown that computer-supported collaborative learning (CSCL) can significantly accelerate human skill formation. The expanding capabilities of LLMs promises to open up new opportunities for advanced learning systems to provide customized instructional support for human learners and new ways of simulating work environments in ways that make that learning practicable and applicable. This talk explores potential, recent explorations, and an agenda with caveats as it looks to the future.

Biography

Dr. Carolyn Rosé’s research group’s interdisciplinary work, published in over 350 peer reviewed publications, is represented in the top venues of 5 fields: namely, Language Technologies, Learning Sciences, Cognitive Science, Educational Technology, and Human-Computer Interaction, with awards in 4 of these fields. She is a Past President and Inaugural Fellow of the International Society of the Learning Sciences, Senior member of IEEE, Founding Chair of the International Alliance to Advance Learning in the Digital Era, and Executive Editor (formerly Co-Editor-in-Chief) of the International Journal of Computer-Supported Collaborative Learning. She also serves as a 2020-2021 AAAS Leshner Leadership Institute Fellow for Public Engagement with Science.

Location: LT1
10:00-10:30Tea Break

Outside LT1

10:30-12:00 Session 3A: Symposium (Strand 5)
Location: NIE7-TR701
10:30
Educating the Human in an Age of AI: Agency, Care and Play

ABSTRACT. This symposium explores how reflecting on key features of what it means to be human can provide insights into the uses and abuses of Artificial Intelligence technologies (AI) in the classroom. It explores the skills, values and dispositions that are key to being a human generally, but also more specifically in the context of Singapore society, and examines the risks and possibilities involved in different uses of AI.

Each participant will focus on one unique feature of humanness key to their own teaching philosophy: agency, care and play. We will introduce the theoretical foundations of each feature and discuss its role in the classroom. We will examine existing research on the impact of AI on each feature, including reflecting on our own experiences in our classroom. We will take an interdisciplinary approach by drawing on a range of concepts, evidence and experiences from psychology, philosophy and communications and new media.

10:30-12:00 Session 3B: Presentations (Strand 2)
Location: NIE7-TR702
10:30
Sketching or GenAI? Tool Choice and Group Storyboarding in Design Education

ABSTRACT. This study compares traditional hand-sketching with generative AI (GenAI) image generation as tools for collaborative storyboarding in a student group. Most evidence on GenAI in design is from single users; this study asks what changes when the unit is a group. Eight BA Design students at the Hong Kong Polytechnic University used both tools in counterbalanced within-subject sessions, analysed through Distributed Cognition. Sketching was rated higher than GenAI on five of six measures. GenAI sessions took longer to reach consensus and produced lower decision confidence. The main barrier was the cost of translating visual ideas into English prompts while discussing in Cantonese or Putonghua, not drawing ability. The study argues that prompt-based generation hides the point where an idea becomes an image, breaking the group's shared view of it. It identifies a human-AI-human sequence: sketch to set direction, render with GenAI once the group agrees, then critique before committing.

10:50
Designing AI for Learning, Not Answers: Learner Perceptions of Pedagogically Designed AI Interaction

ABSTRACT. As generative artificial intelligence (GenAI) tools become increasingly associated with rapid answer generation and task completion, students may rely on them for speed and convenience, creating tensions for pedagogically designed AI agents intended to support guided reasoning rather than direct answer provision. However, limited research has examined how students perceive and interact with such agents in authentic higher education settings. This study investigated students' perceptions of pedagogically designed GenAI agents in an undergraduate auditing module at a Singapore higher education institution. Using pre- and post-intervention survey data from 49 final-year students, findings indicate that students generally perceived the agents as useful and supportive of learning, while their verification behaviours reflected an emerging awareness of AI limitations. However, evaluations remained influenced by efficiency-oriented expectations shaped by general-purpose large language models (LLMs). These findings contribute to a growing understanding of how students engage with pedagogically designed AI agents in higher education.

11:10
The AI Design Studio Partner: Cultivating Humanistic Skills in Architectural Education

ABSTRACT. The proliferation of Generative AI (Gen AI) in architecture education often centers on image generation or rapid design exploration. This study details a pedagogical redesign within a first-year foundational architecture design module that leverages Gen AI to cultivate distinctly higher-order cognitive and humanistic skills. Guided by the Brigg’s constructive alignment theory, the intervention embeds Gen AI as a collaborative learning partner across four stages of the design process: empathy mapping, conceptual synthesis, critical evaluation of AI-generated feedback, and verbal defence of design. By transitioning assessments from evaluating the final product to assessing reasoning processes, ethical AI collaboration, and critical judgment, this study demonstrates plausible pathways to cultivate self-directed, future-ready graduates. By making student thinking/process visible while critiquing AI outputs, this practice nurtures critical thinking, ethical reasoning, and nuanced communication skills demanded by professional practice.

11:25
The ecology before the tool: a practitioner inquiry into AI in communication design

ABSTRACT. AI tools in education are often designed for students, not with them. They are layered onto lesson plans without regard for the cognitive work already in motion. Drawing on a two-cycle practitioner inquiry in Studio Project 2, a Year 2 Communication Design subject at Temasek Polytechnic, it advocates that productive AI engagement is not a matter of tool quality but of ecological design within the classroom: intentionally weaving physical externaisation, peer-driven divergence and convergence, and AI-mediated tools in a single learning environment. Across both cycles implemented, this integration of embodied workshop activities produced measurable gains in student confidence, richer conceptual outputs, and deeper tutor discussions. When embedded within a designed ecology, AI functions as one mediating tool among several –– expanding inputs, challenging positions and helping students crystallise their own arguments. Reducing to a chatbot deters learning, regardless how well it is built.

11:40
Generative AI and Creative Arts Pedagogy: A Qualitative Study of Disciplinary Approaches

ABSTRACT. This paper examines how generative AI disrupts the developmental logic of studio-based creative arts education differently across disciplines. Drawing on in-progress qualitative interviews with academic staff at institutions across the Asia-Pacific region, and grounded in Schön’s reflection-in-action and Lave and Wenger’s situated learning, the study develops a comparative cross-disciplinary analysis. Preliminary findings suggest image-based and design fields face more immediate pressure on foundational skill development, while performance and material practices show greater embodied resilience. A provisional framework for critical AI literacy is proposed, defined by judgement, cultural literacy, and conceptual innovation.

10:30-12:00 Session 3C: Presentations (Strand 2)
Location: NIE7-TR703
10:30
Designing a 5E-Based AI Chatbot for Differentiated Biology Self-Study

ABSTRACT. This paper proposes a 5E-based AI chatbot design for differentiated Biology self-study and reports a small pilot implementation with 23 high school students in a pilot lesson on eukaryotic cells. Grounded in differentiated instruction, cognitive scaffolding, and the 5E instructional model, the chatbot was designed to guide learners through Engage, Explore, Explain, Elaborate, and Evaluate phases. The Honey – Mumford model was used only as an illustrative framework for suggesting flexible learning pathway options, rather than as a tool for classifying students into fixed learning styles. A Gemini Pro Gem named 5E-Bio Mentor was examined through a TAM-based student questionnaire and open-ended feedback. Descriptive results indicated positive perceptions of usefulness, ease of use, attitude, and behavioral intention, while qualitative responses highlighted personalization, response style, response time, and contextual continuity as key design issues. The study contributes a pedagogically grounded chatbot design process for responsible AI-supported Biology self-study.

10:50
Learning Language Through Context A Gamified Approach to Cultural Intelligence in Real-World Scenarios

ABSTRACT. Language learning often emphasizes grammar and vocabulary while overlooking contextual and cultural appropriateness in real-life communication. This paper proposes ContexQuest, a gamified language learning system designed to develop linguistic competence and cultural intelligence through scenario-based interaction. The system enables learners to practice language use in real-life situations they may encounter in a target country. Preliminary feedback from a Japanese convenience store prototype indicates that the interaction was perceived as natural, helping users understand contextually appropriate expressions while improving engagement. The proposed approach aims to support practical communication through AI-supported interaction.

11:05
Developing an AI-Enhanced English Medical Interview Course

ABSTRACT. This paper describes the design and early implementation of an AI-enhanced English medical interview course developed through a Japan–Sweden collaborative project. The course combines video-based learning with AI-generated patient interactions to support culturally responsive healthcare communication. AI is used both for content development and simulated patient interviews. Since April 2026, the course has been implemented with 111 third-year medical students at Shimane University. The paper discusses the course design, AI integration, and future directions for healthcare communication education.

11:20
When Does Generative AI Support Human Learning? Lessons from Life Skills Modules in Polytechnic Education

ABSTRACT. Generative AI is present in polytechnic education, yet its place in life skills modules remains unresolved, especially where learning depends on human interaction, judgment, and practice rather than content delivery. Drawing on interviews with 15 module coordinators, this study examines when AI integration supports learning and when it does not. The analysis identifies four recurring pedagogical patterns: purposeful caution, AI as scaffold toward human interaction, cognitive offloading and student resistance, and deliberate non-integration, as well as two structural findings: the discernment gap and equity’s double edge. The study contributes a clearer account of responsible AI integration in life skills education by showing that what matters is not adoption in itself, but whether AI is used to prepare students for life and work with others and technology, bounded so it does not displace the competency being formed, and supported by the judgment, access, and institutional conditions needed to use it well.

11:35
College Students’ Intention to Use AI Agents in Task-Oriented Learning: Integrating TAM with Learner Agency Perspective

ABSTRACT. AI agents are increasingly used to support task-oriented learning in higher education, yet students differ in their intention to use them. Integrating the Technology Acceptance Model (TAM) with a learner agency perspective, this study examined how key abilities (KA) and agentic actions (AA) shaped perceived ease of use (PEU), perceived usefulness (PU), attitude toward using AI agents (ATT), and behavioral intention to use AI agents (BI). The study was conducted in a statistical analysis course, where students used a customized AI agent for method selection, software operations, and statistical output interpretation. Based on 124 valid questionnaire responses and supplementary interviews, PLS-SEM results showed that PEU and PU positively affected ATT, which further predicted BI. KA positively predicted PEU, whereas AA positively predicted PU. PU fully mediated the relationship between AA and PEU. The findings highlight the role of learner agency in AI-agent acceptance.

10:30-12:00 Session 3D: Presentations (Strand 4)
Location: NIE7-TR704
10:30
The Glow-Up Problem: An Exploratory Study of AI Use in Oral Presentations and Learning in the Age of Generative AI

ABSTRACT. Generative AI tools carry significant promise as democratising resources in higher education, particularly for students whose access to quality academic preparation has historically been uneven. Yet whether this promise is being realised in practice remains an open question. This paper reports on an exploratory study examining self-reported AI tool use, citation practices, and presentation scores of 28 undergraduate students at a large private technical university in India. Analysis reveals that while AI tools have contributed to meaningful improvements in the visual and presentational dimensions of student work, their potential to support deeper intellectual engagement remains largely untapped. Patterns of AI use varied meaningfully across student backgrounds, pointing to the importance of deliberate pedagogical scaffolding. The findings suggest that realising the democratising potential of AI requires not only access to tools, but thoughtful instructional design and tools built to scaffold thinking alongside output.

10:50
High Willingness, Low Preparedness, and Weak Integration: AI Literacy and AIGC-Supported Multimodal Picturebook Teaching among Preschool Chinese Language Teachers in Singapore

ABSTRACT. This study examined AI literacy, AIGC tool use, multimodal picturebook teaching, and professional development needs among 261 in-service preschool Chinese language teachers in Singapore. Guided by the Will–Skill–Tool–Pedagogy (WSTP) model, the study used a six-dimensional AI literacy measure and a nine-item multimodal integration index. Teachers showed moderate but uneven AI literacy (M = 5.27/10), with reflective adaptation and self-efficacy significantly higher than knowledge understanding and tool application. Generative chat tools were widely used (83.1%), but education-specific AI tools (18.4%) and routine picturebook-plus-multimodal plans (2.3%) remained uncommon. Exploratory analyses showed that AI literacy, centre support, and training experience were positively associated with multimodal integration, whereas use frequency made no significant independent contribution. The findings identify a contextualized AI literacy structure in which reflection and efficacy precede knowledge and application, and indicate that pedagogically coherent integration depends on transforming personal preparedness and external support into instructional design.

11:10
Factors predicting K-12 teachers' GenAI use frequency and perceived effectiveness

ABSTRACT. This study examined factors associated with K–12 teachers' generative artificial intelligence (GenAI) use frequency and perceived effectiveness of GenAI-supported teaching behaviors. Drawing on a survey of 2,173 in-service K–12 teachers in China, multinomial logistic regression and multiple linear regression were used to analyze the roles of teacher background, instructional context, and AI literacy. Results showed that AI literacy was consistently associated with higher use frequency and was the strongest predictor of perceived effectiveness. Educational attainment, school region, and subject area were also related to use frequency, indicating that routine GenAI use is shaped by both competence and context. Perceived effectiveness showed a more competence-centered pattern, highlighting teachers’ capacity to evaluate, adapt, and integrate AI-generated outputs. The findings underscore the need for AI literacy development and context-sensitive professional support.

11:30
Beyond Cohesive Development: Latent Profiles of Teachers’ AI Literacy Based on the UNESCO Framework

ABSTRACT. This study examines whether teachers’ AI literacy dimensions — attitudes, knowledge, ethics, and skills — develop cohesively or form distinct configurations. Using a 20-item instrument grounded in the UNESCO AI Competency Framework and latent profile analysis of Lithuanian teachers (N = 448), we identified three profiles: a coherent high-competence group, a low-competence group with ethical and knowledge gaps, and a discrepant group where knowledge coexists with low practical skills. Profile membership was predicted by AI training and frequency of use rather than demographic characteristics, suggesting that differentiated, practice-oriented training is needed to bridge the gap between understanding AI and applying it pedagogically.

11:45
The Relationship Between Teachers’ Attitudes Toward Formative Assessment and Intention to Use AI in Formative Assessment

ABSTRACT. This study examines the relationship between teachers’ attitudes toward formative assessment and their intentions to use AI in formative assessment. Grounded in the Theory of Planned Behavior (Ajzen, 1991), it hypothesises that positive attitudes are positively associated with AI use intentions. A cross-sectional survey was employed with a representative sample of Lithuanian general education teachers (N = 448) using probability cluster sampling. Attitudes were measured using Yan and Cheng’s (2015) scale, and a newly developed 16-item TPB-based scale measured AI use intentions. Preliminary results indicate that attitudes toward formative assessment exert a significant indirect effect on AI use intentions, mediated through attitudes toward using AI in formative assessment. The study extends TPB to AI adoption in formative assessment and aims to inform professional development strategies.

10:30-12:00 Session 3E: Presentations (Strands 2 & 3)
Location: NIE7-TR705
10:30
Generative AI as a Metacognitive Partner in Design Learning: A Mixed-Methods Study

ABSTRACT. Generative AI (GenAI) is often feared to encourage cognitive offloading, yet it may instead scaffold the metacognitive agency central to self-regulated learning. This mixed-methods study examined whether a five-day GenAI-supported workshop could activate metacognition in undergraduate design-based problem solving. Ten second-year students alternated between collaborative human–human and individual AI-supported tasks, completing surveys, reflections, and pre/post Metacognitive Awareness Inventory (MAI) measures. Knowledge of cognition rose significantly (dₚ = 0.80, p = .032), with a medium-effect trend in total MAI under a strong ceiling. AI sessions generated far more ideas (28.42 vs. 5.42), higher suggestion adoption, and more prompting time, while feeling less effortful yet more creativity-supporting and psychologically safe. Learners used AI to externalise and structure alternatives, treating prompt revision and reliability checking as novel metacognitive challenges. We propose a "metacognitive scaffold model": GenAI amplifies ideational fluency and self-regulation only within reflective structures, positioning prompt regulation as a new self-regulatory domain.

10:50
GenAI-Enabled Emotional Regulation Support in Second Language Learning: Pre-Service Teachers' Attitudes and Design Considerations

ABSTRACT. This empirical paper examines pre-service teachers' attitudes toward GenAI-enabled emotional regulation support for second language learners. A descriptive survey of 437 pre-service teachers in Coimbatore district used a 25-item attitude scale covering perceived usefulness, pedagogical value, adoption, ethical concerns, and readiness. Results show predominantly moderate overall attitudes, with strong endorsement of pedagogical value but more cautious views on adoption, ethics, and training readiness. The paper translates these findings into design considerations for classroom-ready GenAI emotional scaffolding, including teacher control, privacy-aware interaction, non-clinical support boundaries, transparency cues, and culturally responsive language learning support.

11:10
Manifestations of Agentic Engagement in AI-supported EFL Classrooms: A Systematic Review through HITL Theory

ABSTRACT. As Generative Artificial Intelligence (GenAI) seems to be embedded in English as a Foreign Language (EFL) classrooms, understanding student agency is crucial. Through the lens of Human-in-the-loop (HITL) theory, this systematic review examines how agentic engagement manifests in AI-supported EFL environments, positioning language learning as human-centric. Synthesizing empirical studies from the past 3 years, the findings reveal four key manifestations of agentic engagement: student-led collaboration with AI, student-teacher navigation of AI integration, essential teacher mediation for critical AI use, and student disengagement. These themes demonstrate that AI-supported learning is a relational and negotiated process shaped by student-teacher-AI interactions. Ultimately, this secondary empirical data shows a nuanced understanding of how students exercise agency, offering critical insights for designing purposeful, pedagogically grounded, and student-centric 21st century foreign language learning environments.

11:25
Learning to Innovate with AI: AI Proficiency, Innovation Capability, and Failure Tolerance in Undergraduate Students

ABSTRACT. This study examines whether an AI-integrated design course affects AI proficiency, propensity to innovate, and failure tolerance among 76 first-year university students in Singapore. Paired t-tests showed significant improvement in AI proficiency, with the largest gains in prototyping. Although overall propensity to innovate and failure tolerance did not change, significant improvements in originality, efficiency, and affect subscales suggest increases in novel thinking, systematic problem-solving, and emotional resilience. Findings demonstrate the potential of AI-integrated design courses to foster specific innovation-related competencies and inform AI integration in design curricula.

10:30-12:00 Session 3F: Presentations (Strands 2 & 4)
Location: NIE7-TR708
10:30
Who Judged the Music?

ABSTRACT. Generative AI can now produce plausible musical materials quickly, but plausible output does not by itself evidence musical learning. This conceptual paper asks how higher music educators might assess human judgement when AI participates in creative production. I argue that AI-assisted music assessment should attend not only to what students produce, but to how they hear, compare, diagnose, revise, and justify AI-mediated musical material. Drawing on evaluative judgement, arts assessment validity, aural fluency, and practice-led music pedagogy, the paper develops an assessment-design account of human musical judgement as visible evidence. It positions AI-generated material as a discipline-specific stress test for higher music education: when musical surfaces become easier to generate, assessment must more carefully identify where human listening, revision, contextual reasoning, and artistic accountability occur. The paper contributes a framework for making musical judgement assessable in AI-mediated creative learning.

10:50
Beyond the Answer Engine: AI as Dialogic Partner for Ethical Reasoning Development in Pre-Service Teacher Education

ABSTRACT. This paper examines whether artificial intelligence, when positioned as dialogic partner rather than efficiency tool, can enhance ethical reasoning capabilities in pre-service teachers. Drawing on an empirical study with Queensland pre-service teachers engaging AI to construct ethical case studies, findings demonstrate that dialogic AI engagement widened thinking, surfaced diverse perspectives, and prompted recognition of ethical grey zones overlooked in conventional ethics instruction. Theoretical grounding in ethical resilience and uncertainty theory, alongside qualitative analysis of student reflections and AI interaction patterns, reveals cognitive and affective shifts when AI functions as Socratic interlocutor rather than answer provider. Implications address retention concerns linked to moral distress whilst challenging instrumental conceptions of educational technology.

11:10
AI in Design Education: A Comparative Study of Assessment Approaches in Emerging Technology Subjects

ABSTRACT. This paper examines pedagogical approaches to integrating the use of generative artificial intelligence and agentic AI in studio-based design education through a comparative study of two subjects at Temasek Polytechnic, School of Design. Building on a contextualised Artificial Intelligence Assessment Scale, the study compares ‘Digital Innovation’, which adopts a May Use AI approach, and ‘Media Futures’, which adopts a Must Use AI stance. These differing conditions reflect distinct teaching strategies, balancing domain-centred skill development with AI-centred capabilities. Using comparative analysis, the study investigates how varying levels of AI integration shape curriculum design, assessment practices and learning experiences within studio environments grounded in documentation, critique and reflective practice. Attention is given to how pedagogical intent is translated into assessment conditions and how students engage with AI as part of iterative design processes. The findings provide insights into pedagogical approaches which may support meaningful and coherent integration of AI within design education.

11:25
Academic Identity in the Age of Artificial Intelligence (AI)

ABSTRACT. Drawing on the view that teacher professional identity is dynamic and continually reinterpreted, the paper argues that AI’s impact is not only technological, but also pedagogical and professional. As generative AI becomes capable of generating content and feedback, lecturers must reconsider their distinctive contribution to student learning. While AI may support lesson preparation, personalised learning and administrative efficiency, it may also blur the boundaries of teaching expertise and constrain professional judgement through algorithmic recommendations It invites educators to an important conversation about the identities and roles of polytechnic educators in the age of AI.

10:30-12:00 Session 3G: Presentations (Strands 2 & 6)
Location: NIE7-TR710
10:30
Designing AI as Cognitive Disruption: An ADAPT Platform for Cultivating Adaptive Thinking in Secondary Science Learning

ABSTRACT. As AI rapidly integrates into the field, educators urgently need to reconsider its use and redesign the AI section to trigger students' deep thinking and support cognitive and adaptive thinking, especially in complex situations that are becoming ever more urgent.This paper focuses on how to integrate AI systems into adaptive thinking cultivation in schools.This study has two significant implications: First, at the theoretical level, it seeks to clarify the core dimensions of adaptive thinking in school learning. Second, at the design level, the proposed ADAPT learning platform design will create a learning process that requires students to continuously adjust their thinking through contextual tasks, conflicting information, constraints, and reflective prompts.

10:45
Collaborative Learning with Generative AI in Teacher Education: A Process Mining Approach

ABSTRACT. This case study examines how interaction unfolds when generative AI (GenAI) is integrated into collaborative learning. Focusing on pre-service teachers’ small-group work, we investigate how learners engage in collaborative prompting and negotiation with each other and with GenAI over time. Using fine-grained interaction coding and process mining, we analyze temporal patterns of key activities—prompting, evaluation, and pedagogical discussion—to identify dominant trajectories and the role of GenAI in collaborative pedagogical reasoning. Pedagogical discussion dominated and was closely intertwined with GenAI-related activity, particularly prompting and surface-level evaluation. Process mining revealed iterative cycles of prompting, evaluation, and discussion, with limited use of external sources. Despite prevalent surface-level engagement, results indicate some degree of reflective collaborative meaning-making supported by GenAI.

11:00
Who Are the AI-Literate Students? Uncovering Dimensional Profiles of AI Literacy in Higher Education

ABSTRACT. This study leverages a person-centered approach to identify distinct AI literacy profiles among higher education students in South Korea based on their intra-individual patterns across five AI literacy dimensions. Latent profile analysis of ipsative scores revealed three profiles: Balanced, Information and Communication-oriented, and Ethics-oriented. The Balanced profile demonstrated significantly higher overall AI literacy and prior AI experience than the other profiles, suggesting that fostering balanced development across AI literacy dimensions may support students’ overall AI literacy development.

11:15
PhysicsPlay: Designing Self-Directed Physics Learning Suites With Google’s Antigravity

ABSTRACT. PhysicsPlay is a game-based learning initiative designed for undergraduate Physics that promotes active, self-directed learning. Developed with Google’s Antigravity, the platform generates varied physics problems, provides embedded resources, supports group discussion, and records student interactions. Its design encourages students to explore concepts, diagnose their own misunderstandings, use curated materials, and improve their solutions through feedback. A key feature is its sustainable feedback model, which uses rule-based responses wherever possible and reserves generated AI support for more complex learning difficulties. This project is intended to improve conceptual understanding, strengthen independent learning habits, and provide support for large Physics cohorts in a more scalable way.

11:00-13:00 Open Demo Session

Visit technology-enabled learning spaces in NIE

12:00-13:00Lunch

Outside LT1

13:00-14:00 Keynote 5: Ast/P Farhan Ali

When Answers Are Cheap, Questions Matter More: Curiosity and Learning in AI-Dominant Education

Synopsis

As Generative AI makes answers increasingly cheap and easy to obtain, the challenge in education needs to shift to fostering better questioning. This keynote argues that questioning driven by curiosity is central to effective learning in digital environments. Drawing on large-scale data and our new frameworks, this keynote explores how questioning and curiosity operate and impact outcomes. This effort is crucial given evidence of how current GenAI use may narrow questioning, curiosity, and inquiry. The talk concludes with principles and interventions that can cultivate agency for deeper and broader self-directed questioning.

Biography

Farhan Ali is an Assistant Professor at the National Institute of Education, Nanyang Technological University, Singapore. His research group addresses the key challenge of global declines in motivation in education by understanding and fostering curiosity in learning, self-directedness, and personal agency and impact in technology-rich environments. He leads the development and deployment of several generative AI platforms, such as TeacherGAIA and GAIA21 , that support curiosity, self-directedness, and 21 st Century competencies in over 100 schools with thousands of students. His group has published in international journals such as Computers & Education: AI, Studies in Educational Evaluation and Journal of Youth and Adolescence, and presented papers at the International Conference on Artificial Intelligence in Education and AERA.

Location: LT1
14:00-14:45 Spotlight 5: Dr Mohamed Arif Bin Mohamed

Teaching Beyond the Prompt: Designing AI Experiences That Make Students Think

Synopsis

Generative AI can give students answers very quickly. For educators, the more important question is how we use AI without taking away the thinking that students are supposed to do.

In this session, I will share some of my experiences using AI in engineering education, from automated feedback to Socratic learning support and, more recently, AI agents.

One area of my work looks at automated formative feedback. The aim is not only to give students feedback more quickly, but also to use feedback to encourage them to revisit and improve their work. AI can also be used in a more Socratic way, by asking students to explain their reasoning, justify their choices and think again about their assumptions instead of simply giving them the answer.

I will also share an activity where students build a narrative AI agent to tell the story behind their data science project. Traditionally, students would analyse their data and present their findings themselves. With the AI agent, they can now compare their own interpretation with the story produced by the AI. This gives them something else to question: Did the AI interpret the plots correctly? Did it miss anything? Did it make assumptions that are not supported by the data? Do they agree with the story it tells?

The main idea behind the session is that AI should not simply make things easier for students. It should also create opportunities for them to question, compare, reflect and make judgements for themselves.

Short biography

Dr Arif Mohamed is a Lecturer in the School of Mechanical and Aerospace Engineering at Nanyang Technological University, Singapore. He teaches computational fluid dynamics, computational thinking, data science and AI and aerodynamics, and is interested in how AI can be used meaningfully in teaching and learning.

His recent work includes AI-supported formative feedback, Socratic learning approaches and the use of AI agents in engineering education.

Location: LT1
14:45-16:00Concurrent Tea Break

EMERGENT NIE7A-Level 1

14:45-16:00 Session 4A: Presentations (Strands 5 & 6)
Location: NIE7-TR701
14:45
Derridean Hauntological Thanabots of Heroes in Social Studies Education

ABSTRACT. Teaching national heroes in social studies education has long relied on static representations and even with current forms of generative artificial intelligence has its own limitations. Drawing from Derrida’s hauntology, this paper argues that thanabots — AI systems trained on the verified writings, speeches, and recorded reasoning of historical figures — offer a more grounded medium for reintroducing national heroes into education. We did this by establishing three distinct forms of haunting: passive haunting, where engagement is one-directional; active haunting, where back-and-forth engagement becomes possible but constrained by diffused corpus; and interactive haunting, where a corpus-constrained thanabot enables direct engagement with what a historical figure actually wrote and argued. The thanabot, thus emerges as an interactive medium through which the historical figure’s trace exerts a hauntological effect on the present. Lastly, the paper responded to the potential objections, arguing that these challenges strengthen rather than undermine our argument.

15:05
AI-Based Emotion Detection in Special Education: A Neurodiversity-Informed Critique

ABSTRACT. AI emotion detection shows promise in special education, yet its ethical implications for neurodivergent students remain unexamined. This paper employs Judy Singer's politics of neurodiversity and Nick Walker's neurodiversity paradigm to critically examine AI-based emotion detection as applied to neurodivergent students. The analysis proceeds in three parts. First, it examines how emotion detection is conceptualized and designed to detect an individual’s emotion. Second, it identifies potential issues when applied to neurodivergent students, particularly that AI-based emotion detection relies on facial expressions, vocal patterns, and body movements, reduced to fixed emotional categories that fail to capture neurodivergent complexity. In this sense, the paper advances a three-part critique: political temporality of recognition; decontextualized medicalization; and neurotypical perceptual dominance. Third, it proposes psychedelics, virtual reality, and cognitive enhancement as alternatives, shifting responsibility toward educators and institutions. We conclude that special education must treat neurodivergent students as subjects deserving recognition, not problems requiring correction.

15:25
Making Sense of Generative AI: A Qualitative Study of Students Who Entered Higher Education After ChatGPT

ABSTRACT. The release of ChatGPT in 2022 has created a sensation across the globe for its ability to imitate human capabilities. The students who entered higher education in 2023 are either under pressure to adopt it or curious about experimenting with it. This qualitative descriptive study incorporated phenomenological method to understand how the students are using GAI in higher education using following research questions, For what purposes are students using GAI tools? How do students interpret and understand the GAI tools? What are the students’ concerns about the GAI tools? The data were collected from students of integrated teacher education program, through semi-structured interviews consisting of 30 questions. The data were coded and analysed thematically. Five themes evolved from the data, purpose, awareness, usage, understanding, and concerns. The findings include students use of GAI for assignments, and their concern of inaccuracy in regional language and reading symbols of math and science.

15:40
From Moral Overload to Institutional Support: Reducing University Teachers’ Proof Labour in the Generative Artificial Intelligence (GenAI) Era

ABSTRACT. Generative Artificial Intelligence (GenAI) is increasingly placing university teachers in a dilemma. They are expected to rigorously regulate student academic integrity while simultaneously maximizing their research efficiency in an environment increasingly shaped by GenAI. This role conflict, compounded by rising administrative burdens, could create a state of moral overload in which professional ethics are crowded out by the sheer labour of proving compliance. To reduce the risk of cumulative strain and burnout of teachers, governance should move beyond merely issuing ethical mandates and instead build institutional infrastructures that shield teachers from the escalating burden of proof labour. We therefore aim to construct an analytical framework to explain how the dilemma of moral overload is generated and to propose a three-dimensional governance roadmap as a set of mitigation conditions.

14:45-16:00 Session 4B: Presentations (Strand 2)
Location: NIE7-TR702
14:45
Designing LLM-Based Educational Simulations to Preserve the Learning Task: A Boundary-Structure-Semantics Prompt Framework

ABSTRACT. LLM-based educational simulations can expand access to professional practice, but default model helpfulness can weaken the learning task by disclosing information or resolving conflict before learners have done the work the simulation is meant to elicit. Drawing on simulation-based education, this conceptual paper argues that prompt design for LLM simulations should preserve the learning task rather than maximise conversational realism. We propose the Boundary-Structure-Semantics (BSS) prompt framework, which organises prompt design around operational boundaries, task progression, and contextual and persona grounding. Retrospective design analysis of two author-designed running examples, a medical history-taking simulation and a leadership negotiation simulation, shows how prompt designs kept key outcomes contingent on learner action and made failures diagnosable. The paper contributes a vocabulary for designing and revising LLM-based simulations when helpful model behaviour removes the intended practice.

15:05
Designing an AI-Infused Scaffolded Curriculum to Cultivate AI-Directors: A Design-Based Research Approach for Non-Technical Polytechnic Students

ABSTRACT. Non-technical polytechnic students in Singapore demonstrate limited higher-order engagement with AI, defaulting to passive or shortcut use in the absence of structured curriculum support. This paper presents the curriculum design rationale and two-cycle Design-Based Research framework for an AI-infused scaffolded curriculum developed for the Diploma in Chinese Media & Communication at Ngee Ann Polytechnic. The curriculum positions students as AI-directors: informed agents who exercise judgement in selecting and directing AI tools to co-create outcomes beyond individual capability. Grounded in a multidimensional AI literacy framework, Vygotsky's Zone of Proximal Development, and the PAIR framework for human–AI interaction, the two-phase design employs domain-knowledge-first sequencing and a craft-AI comparison approach to develop the evaluative foundation necessary for higher-order AI engagement. Five theoretically derived design propositions are articulated, with implications for curriculum designers and educators in media and humanities disciplines seeking to cultivate meaningful AI literacy beyond tool adoption.

15:25
Redesigning a Business Case Assignment with AI-Simulated Executive Interviews: Preliminary Findings

ABSTRACT. Professional financial analysis requires interpreting documents and eliciting managerial perspectives, yet case assignments often provide students with curated materials. We redesigned a graduate finance assignment in which students gathered information by interviewing three role-differentiated AI executives grounded in one anonymised IPO prospectus. Students received feedback between interviews before making investment recommendation. In a pilot implementation (n = 36), the AI executives remained grounded in the source document, did not invent financial information, and maintained role-specific perspectives. Students engaged positively with the activity and reported increased readiness to interview management. However, confidence in challenging vague or overly optimistic answers did not increase. The study offers a feasible model for AI-mediated stakeholder inquiry in document-rich professional learning.

15:40
Process Development of Children L2 e-Books using Generative AI Tools

ABSTRACT. Innovations in generative Artificial Intelligence (genAI) have levelled the field for development of educational materials that were traditionally reliant on creative talent. In this study, we leverage modern AI tools Gemini and Seedance to create a Mandarin e-book for young children. Closely following design principles as outlined by the Cognitive Theory of Multimedia Learning, targeted prompt engineering was employed at the narrative creation, image generation, and multimedia enhancement inclusion stages. Reviews of the output generated found that AI was still lacking in following details for continuity across pages, requiring careful reengineering and revision of prompts to achieve the desired goal. Additions of multimedia enhancements in the form of narration, sound, and motion effects was also guided by feedback in multiple iterations. These processes outline the possibilities and limitations of genAI for creating e-books for children and the collaborative roles performed by human and machine to achieve design goals.

14:45-16:00 Session 4C: Presentations (Strands 2 & 4)
Location: NIE7-TR703
14:45
Collaborative AI-Assisted Translation of Pragmatic Implicatures in Tamil Literary Discourse: Implications for Language Education
PRESENTER: Alisha O

ABSTRACT. AI-assisted translation tools often produce fluent Tamil-to-English output while weakening pragmatic implicatures in literary discourse. Using Vairamuthu's Karuvachi Kaviyam as a case text, this paper proposes an education-oriented framework for reviewing AI draft translations of high-context Tamil literary language. The framework combines Gricean pragmatics, translation quality assessment, and human-AI trust calibration to identify implicature triggers, common AI error modes, and review actions. It contributes a pragmatic fidelity rubric, a taxonomy of implicature triggers, and classroom activities for translation education. The paper argues that AI should be used as a draft generator and discussion partner, not as an authority on literary meaning.

15:05
Integrating GenAI feedback into process writing: Teachers’ and students’ experiences

ABSTRACT. This design-based research examined how English Language teachers designed and enacted GenAI feedback practices and how students engaged with GenAI feedback. Collaborating with secondary teachers in a professional learning team, we explored the integration of GenAI feedback across pre-writing, while-writing, and post-writing stages.

Data sources included planning discussion transcripts, lesson observations, student surveys and focus group discussions. Analysis employed the Pedagogically Productive Talk framework and AI-assisted coding to identify patterns in teacher discourse and student responses.

Findings revealed that teachers transformed from feedback providers to sense-making guides through inquiry focused on students’ needs. Teachers discovered design principles to reduce extraneous cognitive load and promote metacognition. Students demonstrated three levels of feedback engagement: passive (uncritical acceptance), active (contextual evaluation) and strategic (orchestrated integration).

The study contributes a framework for responsibility sharing of GenAI feedback in process writing and supporting feedback engagement, with implications for professional learning design and differentiated pedagogical approaches.

15:25
Designing AI Scaffolds for Tamil Idiom Learning Through Project-Based Learning and Design Thinking: A Design-Based Research Study

ABSTRACT. This design-based research (DBR) study investigated how artificial intelligence (AI), project-based learning (PBL), and Design Thinking (DT) can be integrated to support Tamil idiom learning in a Singapore secondary school. An AI-enhanced PBL intervention was iteratively designed and refined through a precursor iteration and three design cycles. Students explored, applied, and communicated the meanings of idioms through digital storytelling and authentic artefact creation. Data were collected from student artefacts, surveys, reflections, classroom observations, and design documentation. Findings indicate that PBL and digital storytelling supported deeper comprehension of idioms, whereas DT required substantial scaffolding. The study also identified limitations of generic AI tools, including inaccurate feedback and risks of over-reliance. In response, teacher-designed AI tools were developed to provide curriculum-aligned support while preserving learner agency. The study contributes design principles for integrating AI, PBL, and DT in Mother Tongue language education and highlights the importance of human-centred AI scaffolds.

15:45
Student-Curated AI Tutors for Flipped Classrooms to Enhancing Pre-Class Preparation and Self-Directed Learning

ABSTRACT. Flipped classroom effectiveness is often constrained by limited student engagement in pre-class preparation and increasing demands for continuous academic support. Although AI provides scalable 24/7 assistance, lecturer-developed AI systems face challenges in technical expertise, maintenance, and institutional approval. This study proposes a student-centered approach in which learners build their own AI tutors using structured prompts and lecturer-provided materials. In the flipped classroom, students first complete AI-supported pre-class learning, followed by an in-class readiness quiz to identify misconceptions. Lecturers then address gaps before facilitating active learning activities focused on application. Students subsequently complete additional quiz attempts independently, targeting scores of ≥90%, aligned with mastery learning (Bloom, 1968) and self-regulated learning (Zimmerman, 2002). Conducted across three engineering classes, this ongoing study evaluates engagement and learning outcomes through surveys and performance comparisons. The approach provides a scalable, sustainable model for AI integration that promotes deeper learning and learner autonomy.

14:45-16:00 Session 4D: Presentations (Strand 4)
Location: NIE7-TR704
14:45
From Bloom’s Ladder to Bloom’s Loop: AI compresses the middle stages, making process-based assessment necessary

ABSTRACT. This paper proposes and tests an AI-enabled Bloom’s loop as a process-based assessment model that makes students’ learning journeys visible while examining its effects on student learning experience, engagement, anxiety, and grade differentiation. Bloom’s taxonomy should be reframed as a compressed learning loop whereby GenAI accelerates its middle levels, affording learners to apply, analyze, evaluate faster with more polished outputs. This paper posits educators are becoming agency leaders to guide student learning, and a changed assessment design is needed to progress beyond final-output-focused using Bloom's action words and make visible the learning journey. The implication is that new assessment models are needed to maintain fairness, validity, and differentiation. A longitudinal comparison research method is proposed to test this AI-enabled learning and assessment model. The likely outcome should see more homogenized grades while also causing some transition anxiety, but richer engagement and learning experience are expected.

15:00
Weekly Quizzes as AI-Supported Responsive Teaching in Physics

ABSTRACT. This paper examines the use of frequent low-stakes quizzes as an AI-supported strategy for retrieval practice and responsive teaching in A-Level physics. Physics learning is cognitively demanding because the concepts are intricate and students must transfer fundamental principles to novel scenarios rather than merely recall formulae or reproduce worked solutions. However, many students rely on cramming, memorising solution patterns, and “plug-and-chug” approaches, which can create an illusion of mastery. To address this, conceptual quizzes were administered every one or two weeks based on earlier topics and commonly observed misconceptions. Flexible deadlines allowed students to revise, discuss, and clarify ideas before submission, keeping the focus on learning rather than performance. AI-supported analysis of submission times, item-level response patterns such as discrimination indices provided actionable insights into students’ study habits and conceptual understanding. The findings suggest that weekly quizzes can function as diagnostic tools, retrieval opportunities, and prompts for better learning habits.

15:15
Teacher-Guided Generative AI for Custom Middle-School Physics Assessment Generation: A Singapore School-Based Design Study

ABSTRACT. Most AI tools that generate physics questions give teachers little control over the output. This study tackles that problem with a framework that lets teachers choose the question format, topic, marks, cognitive level, number of calculation steps, and whether the question uses a real-world or textbook context. The system runs on two Singapore government-approved AI platforms, PairChat and AIBots, and is grounded in a knowledge base of GCE O-Level past papers, school examination papers, the national Physics Syllabus, and the C3R assessment framework. In a classroom trial, 37 students completed a practice paper generated through this framework and then filled out a feedback survey. The results were positive with a grand mean of 4.45/5, with 94.6% expressing overall satisfaction and not a single negative open-ended comment. The framework prioritises teacher agency, supports differentiated assessment and models responsible AI use in education.

15:30
DeepSeek Empowering Physics Teaching: An Empirical Study on Cultivating Critical Thinking among JC Students

ABSTRACT. This study proposes a "Dialogue-Questioning-Verification" (DQV) pedagogical model leveraging generative AI (DeepSeek) to cultivate critical thinking in high school physics. By reframing AI "hallucinations" in non-idealized systems (e.g., internal resistance and dynamic torque) as cognitive catalysts, the model guides students from blind trust to evidence-based skepticism. An empirical study demonstrates that the experimental group achieved significantly higher post-test physics reasoning scores than the control group, validating the model's efficacy.

15:45
A TPACK-Informed Conceptual Reference Model for Artificial Intelligence Integration in Higher Education Academic Literacy.

ABSTRACT. The proposed study aims to develop a TPACK-Informed Conceptual Reference Model for integrating Artificial Intelligence (AI) into Higher Education Academic Literacy. This research addresses the urgent need to help lecturers effectively incorporate AI tools into academic literacy education within higher education. Once developed, the model can serve as a tool to support lecturers in the effective use of AI. the main research question is: What are the key components and design principles of a conceptual reference model that supports academic literacy lecturers in integrating AI technologies to enhance students' reading, writing, and critical thinking skills for academic purposes? This is a mixed-method, developmental study that adopts a pragmatic paradigm, with a constructivist ontological stance and a pragmatic epistemological stance. The study will adopt both qualitative and quantitative data analysis approaches.

14:45-16:00 Session 4E: Presentations (Strands 2 & 4)
Location: NIE7-TR705
14:45
From Black Box to Glass Box: Instructor Visibility into Team–AI Collaboration During Design Thinking

ABSTRACT. Process-oriented disciplines such as Design Thinking face a growing blind spot: as student teams turn to Gen AI, instructors can evaluate outputs but remain unable to see how AI shapes the process itself. This exploratory, embedded case study tackles that directly by embedding a shared AI bot in student Telegram group channels across four Design Thinking classes at a polytechnic in Singapore (70 students; 4 instructors). The resulting "glass box" gives instructors real-time visibility into team–AI interactions, surfacing patterns that final submissions never reveal. These include uncritical acceptance of AI responses, uneven participation, and the absence of shared prompting roles. Access to this process layer is already shifting instructor practice, with focus groups pointing to an emerging move from artefact evaluation toward active facilitation of AI-mediated collaboration. Full interaction log and focus group analysis is ongoing.

15:00
Designing a Self-Determination Theory-Based Climate Change Education Program Supported by an AI Chatbot and a Metaverse Learning Environment for Elementary Students

ABSTRACT. This study developed and implemented a climate change education program using an AI chatbot and the ZEP metaverse learning environment based on Self-Determination Theory (SDT). The program was designed using the Self-Determination Theory-Oriented Climate Action Learning (SDT-CAL) model to foster elementary students’ autonomy, competence, and relatedness while promoting climate literacy. Participants were 34 fourth-grade students in South Korea. Climate literacy and self-determination questionnaires were administered before and after the intervention. Results indicated significant improvements in climate literacy, including climate knowledge, perceptions, and actions, as well as significant gains in autonomy, competence, and relatedness. Positive relationships were found between climate literacy and self-determination, suggesting that the satisfaction of psychological needs may support climate learning and action. The findings suggest that AI chatbots and metaverse learning environments can support climate literacy and self-determination in elementary science education.

15:15
When AI tutoring tools go unused: Role of Engagement, Psychological factors and Learning Outcomes in a Business Management Institute

ABSTRACT. This study explores the drivers of student engagement with an in-house developed AI-tutor tool, where the initial rollout of the tool showed limited observed engagement. While existing studies highlight the role of technological drivers (perceived ease of use, usefulness), research on the impact of psychological need satisfaction on student engagement, particularly in business management education, remains limited. The study integrates the Technology Acceptance Model and Self-Determination Theory into a single causal framework, tested through a structured two-phase design with 97 postgraduate students across two units. Findings will clarify the role of perceived ease of use, perceived usefulness, and motivation in driving student engagement, whether targeted nudge interventions by faculty can activate disengaged learners, and whether engagement predicts learning gains. Results will directly inform how AI tutoring tools are designed, deployed, and supported in business programmes globally, offering practical guidance for institutions seeking to move beyond adoption toward meaningful learning impact.

15:30
AI as Co-Teacher in Humanities Education: A QSFI Framework for Trustworthy Human-AI Collaborative Pedagogy
PRESENTER: Alisha O

ABSTRACT. Generative AI is increasingly used in classrooms, but humanities education requires more than efficient content generation. This conceptual paper proposes QSFI, a framework for trustworthy human-AI collaborative pedagogy organised around four interactional moves: questioning, scaffolding, feedback, and interpretation. QSFI treats AI outputs as candidate contributions within a human-led decision-support loop, not as authoritative answers. The framework connects dialogic, sociocultural, and posthuman theories of learning with trust calibration, transparency, and governance requirements. It contributes design requirements, classroom guardrails, prompt templates, and evaluation indicators for AI-supported humanities and language education.

14:45-16:00 Session 4F: Presentations ((Strands 4 & 6)
Location: NIE7-TR708
14:45
Pre-service English Language Students' Perspectives on AI-Generated Texts for Use in the EL Classroom

ABSTRACT. In recent years, GenAI texts have been increasingly discussed as a teaching resource in the language classroom. These texts have complicated the authentic-unauthentic text binary, offering contextually relevant learning material that engage learners whilst simultaneously originating from algorithmic iterative training processes (Chandel, 2024), blending conceptualizations of textual authenticity from traditional and alternative perspectives (Widdowson, 1998; Gilmore, 2007). Hence, definitions of textual authenticity must be refreshed through understanding educators’ beliefs and practices surrounding GenAI texts in the classroom. This study uses a qualitative design via interviews to uncover the various perceptions and practices that pre-service secondary school teachers have regarding (GenAI) texts for language learning. Understanding the perspectives of pre-service language teachers – educators-to-be that will shape GenAI text usage in the classroom – will illuminate how (GenAI) texts are used for language learning, and inform educators on how they can harness (GenAI) texts to create a meaningful language learning experience.

15:00
Towards a Cross-Cultural Framework for Asymmetric Perceptions of Agentic AI in Education

ABSTRACT. As agentic AI systems move beyond reactive tool use to initiate, persist, and adapt within educational processes, understanding how teachers and learners perceive artificial agency has become both theoretically and practically urgent. This conceptual paper challenges the prevailing assumption that perceptions of agentic AI are uniform across roles and cultural contexts. Drawing on Floridi's (2025) characterisation of artificial agency, defined by interactivity, bounded autonomy, and adaptability, it proposes the Cross-Cultural Asymmetric Perception Framework for Agentic AI (CCAP-AgAI). The two-layer framework maps asymmetric perceptual responses across teacher and learner roles and specifies how these asymmetries are mediated, amplified, or moderated by socio-cultural context, drawing on power distance theory, epistemic authority norms, and culturally variable orientations towards self-regulation. Indian higher education serves as a theoretically motivated illustrative case. The framework generates eight testable propositions and a structured research agenda for cross-cultural empirical investigation.

15:15
AIED Under Conscription: Institutional Unplugging and AI-Era Learning in Singapore and South Korea

ABSTRACT. As AI reshapes hiring, classrooms, and skills, this study asks how compulsory military service shapes young men’s access to AI-era learning. Extending Isotani et al.’s (2023) concept of AIED Unplugged from low-resource settings to institutionally imposed contexts, we argue that conscription produces a distinct form of institutionally produced unplugging: a state-mandated, time-bound disconnection from the digital infrastructures on which AI-augmented learning depends. We compare Singapore and South Korea through policy and document analysis and digital ethnography of servicemen’s online discourse. We advance four propositions on how each state’s connectivity and recognition policies mitigate or reproduce post-service inequalities and argue for more intentional AIED design for learners under institutional constraint.

15:30
MyanBridge: A User-Informed Romanization-Based Input Framework for the Myanmar Language Using MyanBridge Transliteration Rules (MTR)

ABSTRACT. Myanmar language remains a low-resource language in multilingual computing and natural language processing (NLP) due to challenges including difficult script input, inconsistent romanization practices, limited datasets and the absence of explicit word spacing. These barriers also make the language difficult for non-native learners and researchers to use digitally. This paper introduces MyanBridge, a user-informed input framework based on MyanBridge Transliteration Rules (MTR), a transliteration-based romanization system derived from informal Latin-script Myanmar typing practices commonly referred to as Myanglish. The framework aims to improve Myanmar language text input accessibility through transliteration rules, phonetic approximations and ambiguity reduction strategies using Latin-script input. The MTR rule set is currently under iterative refinement alongside prototype keyboard development for future multilingual accessibility and Myanmar language NLP applications.

16:00-16:45 Spotlight 6: Mr Felix Goh (Google)

Who’s Driving the Lesson? Keeping Educators and Learners in the Lead with Grounded AI

Synopsis

How do we ensure students can still build an argument when all their AI tabs are closed? Join the Google for Education team to explore how educator-bounded, source-grounded tools like NotebookLM and LearnLM-powered Gemini support—rather than replace—critical thinking. Leave with the "40-Minute Autonomy Workflow" and a practical framework for pedagogical, inclusive, and ethical classroom AI.

Location: LT1
16:45-17:30 Spotlight 7: Ast/P Lee Gyeonggeon

Translanguaging Between Human and Machine Languages

Synopsis

Drawing on our endeavor to build a communicative system involving human agents and machine agents, this talk broadens translanguaging as a lens for understanding human-machine collaboration. It considers computational thinking, prompt engineering, and harness engineering as translanguaging practices, which bridge natural languages and programming languages to exchange meanings, express intentions, and coordinate actions. By appropriating the concept of translanguaging from the education field, this talk discusses the implications for how learners can best utilize resources from both linguistic worlds to collaborate with artificial intelligence.

Biography

Gyeonggeon Lee is an Assistant Professor in the Natural Sciences and Science Education Group at the National Institute of Education (NIE), Nanyang Technological University (NTU), Singapore. He earned his Ph.D. in Science Education from Seoul National University in South Korea in 2023, where he also received his B.Sc. in Chemistry Education and B.Sc. in Computer Science and Engineering in 2016. Before joining NTU, he worked as a Postdoctoral Research Associate at the AI4STEM Education Center at the University of Georgia, USA, from 2023 to 2024.

His recent research focuses on integrating artificial intelligence (AI) into science education, encompassing automatic assessment, human-AI collaboration in learning, and lab safety management. His previous research publications have covered a wide range of topics, including the history and philosophy of science education, curriculum studies, educational technology, and electrochemistry. Building on his diverse research experiences and the interdisciplinarity enabled by AI, he aims to establish multimodal science learning to fulfill whole-person education.

Location: LT1