The Data Exchange with Ben Lorica How to Build AI Copilots That Teach Rather Than Automate
10 snips
Dec 11, 2025 
Stefania Druga, an independent researcher and former Google DeepMind scientist, delves into creating AI tools for young learners. She shares insights on how children's natural curiosity informs better AI design. Stefania champions the Socratic method for teaching, highlighting her work on Cognimates as a supportive learning copilot. They discuss the importance of multimodal interfaces and the challenges of current AI education tools. Listen in as she reveals how real-time apps like MathMind address misconceptions in math, pushing for innovative solutions in AI education.
AI Snips
Chapters
Books
Transcript
Episode notes
Kids Build Quirky, Unexpected Models
- Stefania built Cognimates so kids could train custom image and text models and use them in Scratch-like projects.
- Kids created quirky, unexpected models (e.g., weird hairlines, backhanded compliments) that surprised adult designers.
Give Users A Knob To Control Delegation
- Tools lack an adjustable agency knob to let users control delegation to AI versus staying in the driver's seat.
- Stefania argues for Socratic, question-driven copilots that teach and surface blind spots instead of automating tasks.
Ask Questions Before Pushing Solutions
- Design copilots to ask clarifying questions first and only give direct guidance after users remain stuck.
- Use voice and image inputs for accessibility and to better fit children's workflows.
