I’m excited to share our new preprint, “A Functional Taxonomy of Intelligent Systems in Education,” co-authored with Dorien Herremans. Dorien is an Associate Professor at the Singapore University of Technology and Design (SUTD). She was listed among the world’s top 2% of scientists for 2024 in Artificial Intelligence & Image Processing.
We began working on this paper about a year ago, just as I was starting my PhD. Sharing it now feels especially meaningful. It also marks the first year of my artificial intelligence in education (AIED) journey at the Singapore University of Technology and Design (SUTD).
Why use a functional taxonomy of intelligent systems?
Intelligent systems in education are often described by the algorithms or technologies they use. However, our structured integrative review starts from a different question: What pedagogical function does the system perform?
This function-first perspective groups educational AI by its role in teaching and learning rather than by the technical method under the hood. As a result, it offers a shared way to understand a fast-moving field even as individual models, platforms, and interfaces continue to change.
The project was inspired by A Functional Taxonomy of Music Generation Systems, a paper Dorien co-authored almost a decade ago, drawing on work from her own PhD journey. In turn, that study organised intelligent music-generation systems according to their intended functions. A decade later, we revisit the same function-first idea in a different domain: education.
What the functional taxonomy of intelligent systems covers
Our functional taxonomy of intelligent systems in education synthesises eight pedagogical families and traces how data moves among them:
- Adaptive tutoring — personalising guidance and support to learners’ needs.
- Assessment and feedback — evaluating progress and returning information that can guide next steps.
- Student modelling — representing learners’ knowledge, skills, behaviour, and other relevant states.
- Learning analytics — turning educational data into insights for learners, teachers, and institutions.
- Collaboration support — helping people learn, solve problems, and create knowledge together.
- Content generation — producing or adapting learning materials and activities.
- Game-based learning — using intelligent support within playful and interactive learning experiences.
- Cross-function integration — connecting functions, data, and services across the wider learning environment.
Together, these pedagogical functions make the relationships among systems easier to see. For example, adaptive tutoring may depend on student modelling; assessment can feed learning analytics; content generation can support tutoring, collaboration, or game-based learning. The taxonomy therefore describes more than a list of tools—it shows how educational intelligence can work as a connected whole.
From stand-alone tools to educational ecosystems
One of the paper’s central observations is that the field is moving from stand-alone tools towards interoperable educational ecosystems. As functions become connected, researchers and designers need to consider how data, decisions, and responsibilities flow across the system—not only whether one component performs well in isolation.
Importantly, the review also highlights recurring challenges for artificial intelligence in education: long-term adaptivity, generative reliability, fairness and privacy, teacher–AI collaboration, and system-level evaluation. These issues become increasingly important when intelligent systems influence several parts of the learning process at once.
Moreover, this ecosystem view connects with my broader work on Machine Pedagogical Intelligence and AI in education: the value of an intelligent system should ultimately be understood in relation to the teaching and learning functions it supports.
A meaningful first-year PhD milestone
This paper marks a special milestone in my PhD journey. The work began as I entered the field of AI in education, and sharing it now coincides with the completion of my first year at SUTD.
I’m deeply grateful to Dorien for the inspiration, mentorship, and collaboration that brought this work to life.
Read the paper
Read “A Functional Taxonomy of Intelligent Systems in Education” on SSRN
Permanent DOI: https://doi.org/10.2139/ssrn.7183660
Kadir, Nizam and Herremans, Dorien. “A Functional Taxonomy of Intelligent Systems in Education.” SSRN preprint, 2026. DOI: 10.2139/ssrn.7183660.
Ultimately, if you work in artificial intelligence, education, learning analytics, EdTech, or intelligent systems, I hope this taxonomy offers a useful way to understand the field through the pedagogical functions these systems perform.
Leave a Reply