Kennisbank

AI in education: which tasks are shifting?

Educational institutions combine two types of work that have little to do with each other: teaching and guidance on the one hand, and administration, planning and accountability on the other. The largest part of the non-teaching hours goes into scheduling, testing and grading, reporting to parents and inspectorates, and communication with pupils, students and parents. In addition, there is a layer of policy and quality assurance that is strongly tied to laws and regulations, from the education inspectorate to examination regulations.

Three circumstances determine how that work can be divided between AI and people. First, the degree of personal contact: guiding pupils and students requires assessing someone's development, motivation and wellbeing, something that cannot be captured in a system. Second, the regulatory pressure: exams, diplomas and student records are bound by strict procedures, which means automated output always needs a documented review moment. Third, the fragmentation of systems: many schools work with a student information system, a scheduling package, a communication tool and a separate environment for testing, which often do not connect seamlessly. That fragmentation determines how much time is still needed to transfer data between systems, even when AI can already handle the substantive work.

AI can take over the task

For a number of tasks, the outcome is easy to verify and the risk of an error is limited, which means AI can do the entire job. Think of generating draft schedules based on fixed rules, summarizing course material into practice questions, drafting standard letters to parents, and transcribing and summarizing meetings. Preparing report texts based on grades and attendance, and answering frequently asked questions about enrollment, schedules or exam dates, also fall into this category. That last type of question overlaps with what which work can AI take over in customer service describes: recognizable questions with a fixed answer lend themselves well to full automation.

Partly, with human oversight

A large part of educational work falls into the category where AI produces a draft and a human assesses, adjusts or rejects it. Grading open questions and essays is the clearest example here: AI can generate an initial assessment and feedback, but a teacher must approve or correct it, especially where the grade counts toward a diploma. The same applies to drafting individual development plans, where AI can flag patterns in data but the interpretation of someone's situation remains human work. AI can also provide starting points when preparing conversations with parents or flagging possible learning delays, while the interpretation and communication remain with the professional. Financial and administrative tasks around, for example, grant applications or budgets follow a similar division to what is described on which work can AI take over in the finance department: AI provides the number-crunching and initial analysis, a human remains responsible for the final decision.

What remains human work

The core of education, the actual teaching, guiding and assessing of people in their development, remains human work. This also applies to final assessments that determine a diploma or progression, to conversations about care, behavior or wellbeing of pupils, and to the pedagogical relationship needed to motivate or redirect someone. Policy choices about educational quality, examination and dealing with the education inspectorate require a responsible official who can explain and defend the outcome. Where this touches on personnel decisions, such as the deployment of teachers or support staff, separate legal requirements apply that are independent of what AI can technically take over.

How this relates to other sectors

The division in education shows similarities with other sectors where personal contact and regulation come together. In what can AI take over in healthcare, the same question arises as to how far automation may go for tasks that affect people's safety or wellbeing, while administrative processes can indeed be fully automated. Institutions that also purchase or supply goods, such as when acquiring learning materials, will also recognize elements from what can AI take over in wholesale, where ordering processes and inventory management lend themselves well to automation as long as the systems connect to each other.

What you can do now

The division between these three categories differs strongly per school, per role and per system used. A general estimate says little about the situation at a specific location. Anyone who wants to know which part of their own tasks is already eligible for automation today can fill out the free quickscan from ftetoai: twelve questions, no account needed, with an indication of what part of the hours in your own profile can currently be taken over by AI. The full work scan, which goes deeper into specific roles and processes, is still under construction and is not offered here as a ready-made solution.