Evidence-grounded conversational knowledge exploration
Ask-Julia
A conversational interface designed around retrieval-grounded answers, accessible evidence and structured follow-up.
Explore system →Selected work
Evidence from teaching and applied practice
Selected work across AI-literacy curriculum, assessment, adaptive expertise and educational AI systems.
Curriculum & teaching
Nanyang Technological University
AI Literacy in the NTU Core Curriculum
2024–Present
- Scale
- 1,700+ students
- Delivery
- 42 tutorial classes
- Focus
- Judgement with AI
A case study in integrating grounded research and agentic AI into interdisciplinary undergraduate learning.
The design moves students from foundational readiness to facilitated inquiry and applied mastery. Source libraries, agent-building workflows and human-intervention records make evidence, reasoning and responsibility visible throughout the learning process.
View case study →AI literacyInquiry-led learningHuman judgement
Research & public engagement
Ideas tested in public
Adaptive Expertise for Human–AI Work
Explore the framework →A framework for rethinking job readiness when increasingly capable AI systems can perform routine cognitive work. It places greater emphasis on problem framing, evidence evaluation, exception handling, interdisciplinary judgement and effective human–AI working arrangements.
AI Adoption vs AI Advantage
Read the perspective →A human-centred perspective on organisational AI adoption. As advanced AI becomes widely accessible, the harder problem is redesigning work, building complementary human capability and deciding where automation, augmentation and human judgement each create the most value.
Applied educational AI
Tools as inspectable teaching practice
Each prototype asks a practical question: what should AI do, what should the human continue to do, and how should the interface shape that division of cognitive work?
Structured multi-perspective argument analysis
Critical Thinker
A reasoning environment that makes critique more inspectable and reduces premature acceptance of fluent conclusions.
Explore system →AI-supported parallel thinking for groups
Six Thinking Hats
A collaborative workflow separating facts, emotions, risks, benefits, alternatives and process control.
Explore system →AI-supported instructional design
Generative Lesson Planner
A workflow for creating lesson structures and activities while keeping outcomes, context and educator judgement in control.
Explore system →These systems are prototypes rather than validated products. Their purpose is to make assumptions about evidence, reasoning and human responsibility concrete enough to inspect and improve.