Keynotes

Abstract

Artificial intelligence is reshaping how young children acquire language—at home and in classrooms. This keynote synthesizes evidence from meta-analytic, scoping review, and theoretical research to examine what AI can and cannot do for children's second language development, and what this means for children, parents, and teachers.

Research confirms that AI-based interventions significantly improve children's second language speaking, particularly through personalized, conversational feedback difficult to achieve in typical classrooms. Yet effectiveness data alone cannot guide practice. The more consequential question is how AI works alongside the humans children depend on. Drawing on the Dual Scaffolding Framework, this talk argues that children's language outcomes are shaped by the relationship between AI and human scaffolding—coordinated, redundant, or competing. For parents and teachers, this means reimagining their role as essential co-scaffolders whose engagement determines whether technology truly serves children's language growth.

Abstract

Artificial Intelligence (AI) promises to transform education through intelligent tutoring systems, learning analytics, and generative AI. Yet, despite these advances, most AI applications remain focused on improving individual learning. At the same time, many of today's global challenges—from climate change to scientific discovery—require people to think, learn, and innovate collectively. This raises a fundamental question: How can AI support not only individual intelligence, but also collective intelligence?

This presentation argues that the next frontier of AI in education is a shift from personalized learning to education for collective intelligence. Rather than viewing AI simply as a tutor or productivity tool, education should begin to position AI as a collaborative partner that supports dialogue, knowledge building, and collective problem solving. Collective intelligence does not emerge automatically from connecting people and AI; it develops through meaningful dialogue, shared reasoning, and the co-construction of knowledge.

Drawing on recent research in Artificial Intelligence in Education (AIED), the presentation introduces a framework illustrating how AI can support collective intelligence by facilitating group formation, collaborative dialogue, and learning orchestration. It further explores the emergence of Human–AI Teamwork, arguing that education must prepare learners and teachers to work effectively in hybrid human–AI teams. This requires moving beyond AI literacy toward new capabilities, including collaborative reasoning, trust calibration, ethical judgement, critical leadership, and collective knowledge building.

The presentation concludes by proposing education for collective intelligence as a new research agenda for AI in education. The future of AI in education will not be defined by how intelligently AI teaches individuals, but by how effectively humans and AI engage in dialogue, build collective intelligence, and solve complex problems together.

Keywords: Collective Intelligence; Human-AI Teamwork; Generative AI in Education; Collaborative Learning; Educational Technology Design; Dialogic Education

Abstract

Mathematical problem solving is a long-standing challenge at the education–AI intersection. Over seven decades, methods progressed from rule-based heuristics to computational models. Today, two paradigms compete: state-transform, based on formal state transitions, and LLM, based on generative reasoning. This keynote reviews that evolution, then compares the two approaches on cost, explainability, and classroom readiness. It explores which paradigm—or hybrid—best serves next-generation intelligent education, outlining open challenges for researchers and practitioners.