This project will explore using a LLM that simulates a young child (early years and primary aged; 3-12 years old). The direct goal is to generate a new dimension to teaching and stimulate new questions, such as: Why a Child AI is limited in replacing engagement with real children; What are the limits of developing a Child AI (esp. considering robotics). What is required to simulate a child’s voice beyond language limitations (e.g., listening, conceptual barriers, embodied interaction).
This project will enable students to reflect on the benefits and ethical issues of different groups using this type of technology – e.g., designers, policymakers, trainee teachers, whilst also critically reflecting on the possible risks, notably from nefarious uses of child simulations. The project will also encourage critical thinking around ethical complexities – e.g., perhaps Child-Ais would help represent children’s voices more; encourage designers, policymakers etc. to explore children’s perspectives alongside real-world engagement where feasible. We can reflect on where data on children’s voices might be modelled from – e.g., rise in AI speakers in homes; AI toys etc.
In summary, this project is teaching focused, but raises many ethical, theoretical, technical, design and pedagogical questions that are timely and important for research – with potential for students to publish from their experiences.
