On this page

Technologies — the models, architectures, and methods that power AI education systems, and the umbrella concept for the technical layer. Where pedagogy and learning theories concern how teaching and learning happen, and AI ed evaluation concerns whether AI works, the technical strand covers the AI systems (large language models, generative AI, multimodal models, robots) and the techniques used to build, control, and deploy them (prompt engineering, retrieval-augmented generation, reinforcement learning, educational NLP, knowledge graphs, agentic orchestration).

Questions to Consider

  • Teaching well does not require building an LLM, but the model and retrieval choices underneath an AI tool still decide what it can do in your classroom. What is one way the underlying technology of an AI tool might quietly change how your students learn, even if you never see the code?
  • A common assumption is that the model is the whole story — but techniques like retrieval-augmented generation (RAG) and prompt engineering exist precisely to control and ground LLM output. Before reading further, when you ask an AI to 'be more accurate' or 'use this source', what do you think is actually happening under the hood?
  • RAG is described as a core technique for reducing hallucination and improving safety. Why do you think fetching relevant knowledge to 'ground' an AI's answer would matter more for education than for, say, casual chat — and what could go wrong if that grounding fails?
  • Technical choices embody pedagogical assumptions: a tutor built on Socratic prompting reasons with learners, while an answer-generating model may just hand over solutions. Can you recall an AI tool you've used that seemed to 'assume' a particular teaching philosophy — and did that align with how you actually wanted to teach or learn?
  • Beyond raw accuracy, AI systems should be evaluated on reliability, pedagogy, and equity. What headline metric do you suspect most people (including many educators) default to when judging whether an AI tool 'works', and why might that metric hide more than it reveals?
  • Agentic AI is described as shifting AI 'from a prompt-responding tool into a proactive collaborator.' How might a system that initiates and orchestrates multi-step workflows on its own change what you, as an instructor or learner, are responsible for — and who holds it accountable?

Introduction

AI in education runs on a specific technical stack, and understanding it matters for educators and researchers even when they do not build systems themselves — because technical choices shape what AI can and cannot do in the classroom, the risks it carries, and how to evaluate it. The technical layer covers the AI systems, the techniques that adapt and control them, and how it connects to pedagogy, assessment, and evaluation.

AI systems in education

A function-based taxonomy travels further into teacher professional development than a list of model architectures. Jaksic, Ansaf and Massey (2025) organized a one-day workshop for nine high school teachers around six families of AI application — classification, recognition, memory, decision-making, optimization and prediction — each paired with an application the teachers already used, from spam filtering to credit scoring. Teachers arrived split on whether AI was a threat, with 40% saying so and 40% undecided, and every participant reported learning something new. The organizing unit is what the system does, which a teacher can recognize in a classroom tool without knowing how it was trained.

Techniques and methods

  • Prompt engineering. Prompt engineering is how educators and developers shape LLM outputs — the primary mechanism through which offloading and control are enacted in LLM interactions.
  • Retrieval-augmented generation (RAG). RAG grounds LLM outputs in retrieved knowledge, reducing hallucination and improving accuracy — a core technique for safe educational deployment.
  • Reinforcement learning. Reinforcement learning trains agents to optimize behavior over time, used in adaptive systems and game-based learning.
  • Agentic orchestration. Agentic AI systems plan and execute multi-step workflows — often orchestrating multiple specialized agents (see multi-agent systems) — and are reshaping AI from a prompt-responding tool into a proactive collaborator.
  • The educational agent stack lags the frontier. Wang et al. (2026) mapped 474 studies and found GPT-series models and LangChain dominant while governed tool orchestration, persistent memory, and long-horizon planning were largely absent — and only 138 of 474 (29%) drew on educational theory.
  • Model training and adaptation. LLM training and fine-tuning, educational alignment, and small-language-model adaptation make general models education-specific — though the evidence puts retrieval and prompting ahead of training in the decision order, since a well-grounded prompt is cheaper than an adapted model.

How the technical layer connects to the field

The technical strand is inseparable from the field's other themes:

Research attention is concentrated on a narrow slice of the tool ecosystem. Mapping 85 AI tools against the STEM and engineering literature, Terkowsky et al. (2026) found only 36 research-visible at all and conversational AI/LLMs accounting for 73.7% of 927 paper–tool connections, with 23 publications in an established laboratory context.

Implications for AI in education

  • Technical literacy supports critical use: understanding the underlying models and techniques helps educators and learners use AI well and evaluate it critically (see AI literacy).
  • Choose technology by pedagogical intent: the AI system and technique should follow the teaching strategy, not the reverse.
  • Evaluate the technical layer: AI ed evaluation and benchmark research assess AI systems on reliability, pedagogy, and equity, not just headline accuracy.
  • Robots and agents are part of the stack: embodied and agentic systems extend the technical repertoire beyond text — and bring their own design and safety considerations.

Common Misconceptions

"The model is the whole story." Techniques such as retrieval-augmented generation and prompt engineering exist precisely to control and ground an LLM's output, and the architecture around the model often determines whether it is usable in a classroom. Two systems on the same model can differ entirely in reliability because of what surrounds it.

"The field's technical maturity is established." The mapping is thin where it matters: across 474 studies, GPT-series models and LangChain were dominant while governed tool orchestration, persistent memory, and long-horizon planning were largely absent, and only 138 of 474 (29%) drew on educational theory. The stack being used is far narrower than the stack the field needs.

"Technical choices are the technologist's business, not the educator's." You can be an excellent educator without being able to build an LLM and your technical choices still shape what AI can and cannot do in your classroom. Whether a system grounds its answers in course material is a decision with pedagogical consequences that someone has to make.

"A more capable model will resolve the field's gaps." Orchestration, memory and long-horizon planning are architectural rather than capability problems — the documented absence is of systems that govern their own tool use over time, which a better model does not supply on its own.

Connected Concepts

Connected Articles

Embed this page

Copy the code below to embed this page in a learning management system or other website. The embedded version shows only the page content — it hides the site header, navigation, and footer.