In this EdTechnical short, Libby and Owen test a conversational plush toy to understand more about AI-powered toys designed for young children. Recent research from Cambridge shows that preschool-aged children can form rapid emotional connections with social robots like these, even when the responses from the robot are inconsistent.

Children’s experiences with AI toys are shaped by voice and real-time interaction. Could highly responsive, frictionless AI systems in toys influence children’s expectations of human relationships?

Libby and Owen discuss the difference between shared, supervised play and extended solo interaction with the toy, which may be less advisable. As the technology continues to improve, the key challenge becomes how these tools are introduced and used in early childhood environments.

Links:

• BBC Article: AI toys for children misread emotions and respond inappropriately, researchers warn
• Cambridge study on AI toys in early childhood
• AI chatbots and the “empathy gap” in children

Join us on social media:

• BOLD (@BOLD_insights), Libby Hills (@Libbylhhills) and Owen Henkel (@owen_henkel)
• Listen to all episodes of Ed-Technical here: https://bold.expert/ed-technical
• Subscribe to BOLD’s newsletter: https://bold.expert/newsletter
• Stay up to date with all the latest research on child development and learning: https://bold.expert

Credits: Sarah Myles for production support; Josie Hills for graphic design; Anabel Altenburg for content production.

In this episode of EdTechnical, Libby and Owen speak with Panos Ipeirotis, Professor at NYU Stern School of Business, about his experiment using AI to run oral exams in university courses. As generative AI makes it easier for students to outsource written assignments, educators are asking whether traditional take-home assessments still measure real understanding.

Panos introduced AI-mediated oral assessments after noticing a mismatch between high-quality written submissions and weak classroom discussion. In the new system, students answer questions from a voice agent that probes their understanding of the material and their own work.

Panos tells Libby and Owen how the exams work, including an AI “council” of language models that evaluates transcripts and produces detailed feedback. What does this approach reveal about the future of assessment? Could AI make oral exams scalable in higher education, and even improve fairness and grading consistency?

Links:

• Panos Ipeirotis – NYU Stern Faculty Profile
• NYU Professor Uses AI-Run Oral Exams to “Fight Fire with Fire”
• Article: The case for oral assessment in the age of AI

Guest Bio

Panos Ipeirotis is a Professor of Information, Operations and Management Sciences at NYU Stern School of Business. His research focuses on data science, AI, and human-AI collaboration. In addition to his academic work, he experiments with practical applications of AI in education, including new models of assessment that combine oral exams with AI-based evaluation.

Join us on social media:

• BOLD (@BOLD_insights), Libby Hills (@Libbylhhills) and Owen Henkel (@owen_henkel)
• Listen to all episodes of Ed-Technical here: https://bold.expert/ed-technical
• Subscribe to BOLD’s newsletter: https://bold.expert/newsletter
• Stay up to date with all the latest research on child development and learning: https://bold.expert

Credits: Sarah Myles for production support; Josie Hills for graphic design; Anabel Altenburg for content production.

In this EdTechnical Short, Libby and Owen explore how AI might reshape teaching through the lens of the “weakest link” theory from economics. They discuss the possibility of full job replacement, partial task automation, and productivity gains for teachers. Automation often shifts the composition of work rather than eliminating roles, as with bank tellers and radiologists.

In schools, planning, grading, diagnosing student understanding, classroom management, and relationship-building are tightly interconnected. Automating one component may reallocate time, but complexity is not neatly reduced.

AI can already perform isolated teaching tasks. What happens to the education system when those tasks are embedded in a deeply relational profession?

Links:

• Michael Kremer (1993), The O-Ring Theory of Economic Development
• David Autor (2015), “Why Are There Still So Many Jobs? The History and Future of Workplace Automation”

Join us on social media:

• BOLD (@BOLD_insights), Libby Hills (@Libbylhhills) and Owen Henkel (@owen_henkel)
• Listen to all episodes of Ed-Technical here: https://bold.expert/ed-technical
• Subscribe to BOLD’s newsletter: https://bold.expert/newsletter
• Stay up to date with all the latest research on child development and learning: https://bold.expert

Credits: Sarah Myles for production support; Josie Hills for graphic design; Anabel Altenburg for content production.