All research initiatives
Research Initiative 06
AI in Education
We study how students learn with AI and how teachers use it to support instruction, feedback, and assessment.
We examine whether AI-assisted activities help students understand concepts, think critically, and apply what they have learned to a new task on their own.
Projects
Each entry is labelled with its current stage. Select an area or a study to read more.
4 entries
01
How reliably can AI explain heritage sites, correct misconceptions and write tourism assessment questions?
Earlier proposal
- Scope
- Evaluate AI-generated tourism examination questions for factual accuracy, correct answer keys, ambiguous wording, and fidelity to the source material.
Earlier proposal
- Scope
- Assess the historical and cultural accuracy of explanations that large language models generate about the Wonders of the World.
Incorporated into Idea 4
- Scope
- Compare standard model responses with responses produced using live web search, a fixed set of authoritative sources, or a prompt to verify the answer. This comparison is now part of the multilingual heritage misconceptions benchmark.
Current preferred study
- Scope
- This benchmark will test whether ChatGPT, Gemini, and Claude recognise and correct mistaken premises in questions about heritage sites. It will use authentic heritage misconceptions and misleading questions in English, Hindi, and Malayalam to compare standard responses, live web search, responses grounded in a fixed set of authoritative sources, and prompts that ask the model to verify its answer.



