This talk will discuss the role of creative expression in designing experiences for K-12 learners in online and in-person learning for computational thinking and AI literacy.  It will review several key educational media works from Dr. Magerko's Expressive Machinery Lab at Georgia Tech that intertwine creative expression with technical learning objectives, including EarSketch, an online CS learning environment (co-founded by Drs. Magerko and Freeman) with over 1.5 million active users that uses music making as a way to engage students in computational thinking, and LuminAI, a prototype exhibit where students learn about ML while interacting with an AI improvisational dancer.  This talk will discuss the role of creative expression and embodiment in the Expressive Machinery Lab's collaborations on AI literacy interventions for public spaces, such as LuminAI, and the challenges/opportunities in mapping those benefits to online educational environments.

Much of education involves helping people learn by providing timely, on-point explanations of unfamiliar or difficult topics.   The rise of Generate AI provides a powerful new set of tools for automating explanations, but as the community has rapidly discovered, these tools are still not not usable out-of-the-box for many end-to-end tasks. 

In this talk, I argue that a combination of careful user interface design techniques are necessary (but probably not sufficient) to corral the sometimes unpredictable behavior of LLMs.   I will describe some general principles for how to do this, as well as illustrate examples from my recent work with various co-authors.