There is a part of teaching that is rarely seen by outsiders. It's not when standing in front of the class, but the night before: preparing materials, thinking of suitable activities, creating questions, and then grading each participant's work one by one. This stack of tasks works quietly and often takes more time than the teaching hours themselves.
Recently, more educators have started to involve AI to lighten that load. Not to replace the role of teachers, but to speed up the repetitive and time-consuming parts.
The Hidden Burden
An educator who handles several classes does not just prepare one set of materials. They prepare many versions for different topics and levels. After the materials, there are questions that must be fair and truly measure understanding. There are rubrics to ensure consistent assessment. There are worksheets, and ideally, feedback that is personal for each participant. This administrative work is attached to the profession and is usually done outside of hours, after the children are asleep or during breaks.
What Has Started to Be Delegated
The data is quite clear about the direction it is heading. A Gallup survey recorded that more than 60 percent of public school teachers in the United States used AI tools throughout the 2024 to 2025 school year, with about one-third using them weekly. The most common uses are not dramatic: creating lesson plans (about 38 percent), summarizing information, and producing materials such as questions and assignments (about 37 percent). A similar pattern is slowly emerging in teachers' rooms in Indonesia, especially since tools like Claude, ChatGPT, Gemini, NotebookLM, and Canva have become more accessible. Each has different strengths, and therein lies the challenge: knowing when to use which.
The Line That Must Not Be Crossed
This is where the distinction lies. AI is good for drafting, providing variations, and structuring. But good questions are not just numerous; they must measure the right things. Rubrics should not only sound neat; they must map the learning objectives set by the educator. And useful feedback still requires context that only someone who knows their class can understand. The quality of AI output heavily depends on how the request is framed: what role is assigned, for which audience, with what goals and limitations. Without that framework, the results are generic. With that framework, AI transforms from a toy into a truly helpful assistant.
Tools, Not Replacements
Interestingly, as more teachers use AI, it becomes increasingly clear that human assessment is becoming even more important. Machines can write ten versions of questions in seconds, but deciding which version is appropriate for the class remains the educator's job. Technology provides speed. Understanding the students, what makes a concept finally stick, is something that cannot be delegated to anyone.
References:
- Gallup – Teachers and AI: Adoption in U.S. K-12 Schools (2025) → gallup.com
- Microsoft Education – AI in Education Report (2025) → microsoft.com
- UNESCO – Survey on Institutional Guidance for AI Use (2025) → unesco.org