Over the past 12 years, we have been writing a regular column in TechTrends, broadly focused on “Rethinking Creativity and Technology in Education.” More recently, we’ve focused our attention on the role of GenAI in teaching, learning and creativity.

In our previous article “Control vs. Agency: Exploring the History of AI in Education”, we examined the historical tensions between control and agency in AI-driven educational technologies. We traced how early conceptual frameworks continue to shape current debates by exploring two contrasting visions—John Anderson’s structured cognitive tutoring systems and Seymour Papert’s constructionist approach emphasizing the creative agency of learners.

Our newest piece, co-authored with Danah Henriksen and Jim Dunnigan, extends this historical exploration by focusing on a central divide in AI research: symbolic versus neural approaches to machine intelligence. Though technical on the surface, this tension reflects deeper questions about learning and intelligence that resonate strongly in today’s educational debates.

What makes these paired articles particularly compelling is how they illuminate different dimensions of the same historical trajectory. Interestingly, Seymour Papert emerges as a fascinating figure connecting these narratives. The same champion of student agency and constructionist learning also co-authored a devastating mathematical critique (with Marvin Minsky) that stalled research into and the development of digital neural networks for decades, until their recent resurgence.

This history offers important context for today’s debates about AI in education. Current large language models, built on neural architectures, don’t encode knowledge through explicit rules but learn patterns from vast amounts of data. Their strengths and limitations tell us something essential about learning itself—both its statistical nature and the importance of embodied experience, which these systems lack.

We believe that these two articles, taken together, provide important historical context to understand the intertwined nature of the development of AI, human psychology and education. Citation and link to article given below.

Mishra, P., Henriksen, D., & Dunnigan, J. (2025). From symbols to statistics: The parallel histories of machine and human learning. TechTrends. https://doi.org/10.1007/s11528-025-01083-z