Every conversation about AI in education eventually lands on the same question: "Will it replace teachers?" The answer is no. But that's the wrong question.
The right question is: what parts of teaching are currently done by humans that shouldn't be? Not because they're unimportant, but because they're mechanical, repetitive, and could be done better by a machine — freeing teachers to do the things only humans can do.
What teachers actually spend time on
Research by the OECD's TALIS survey consistently finds that secondary school teachers spend less than half their total working time on actual teaching. The rest goes to:
- Creating and grading assessments
- Administrative tasks
- Lesson planning and material preparation
- Individual feedback on student work
- Tracking student progress
Much of this is work that AI can meaningfully assist with — not by doing it badly and approximately, but by doing the mechanical parts well enough that the teacher can focus on the judgment, creativity, and human connection that actually makes education work.
Where AI genuinely helps
Assessment generation. Creating good test questions is time-consuming and surprisingly difficult. AI can generate questions at specific difficulty levels, aligned to specific learning objectives, in seconds. The teacher's role shifts from creating questions to curating and quality-checking them — a much more efficient use of their expertise.
This is something we built into Edukamentas. Our AI generates practice tests based on the actual content students are studying, targeting the specific concepts they need to practice. Teachers can review and adjust the questions, but they don't have to start from scratch.
Personalized practice. In a class of 30 students, each student has different gaps in their knowledge. Providing truly personalized practice is impossible for a single teacher to do manually. AI can analyze each student's performance and surface the right practice problems at the right difficulty level.
Content adaptation. Students learn at different paces and respond to different explanations. AI can rephrase explanations, provide additional examples, or adjust the reading level of material to match individual student needs.
Where AI shouldn't go
There are parts of education where AI has no business being the primary actor:
Motivation and mentorship. A student struggling with confidence doesn't need an algorithm. They need a human who sees them, believes in them, and can adjust their approach based on dozens of subtle social cues that no AI can read.
Critical thinking development. AI can help students practice applying knowledge, but the deep work of learning to think critically — questioning assumptions, evaluating evidence, constructing arguments — requires human dialogue and challenge.
Social and emotional learning. Learning to work with others, manage emotions, and navigate conflict are fundamentally human skills that develop through human interaction.
The real shift
The most productive way to think about AI in education isn't replacement — it's redistribution. AI can take on the tasks that are mechanical and repetitive, giving teachers more time for the tasks that are relational and creative.
A teacher who spends two hours creating a test could instead spend those two hours in one-on-one conversations with struggling students. A teacher who spends their evening grading worksheets could instead spend that time designing a project that sparks genuine curiosity.
That's not a future we need to be afraid of. It's one we should be building as fast as we can.
At Uncascade, that's exactly what we're trying to do. Not replacing teachers. Making the parts of their job that matter most — the human parts — possible to prioritize.
References
OECD. (2019). TALIS 2018 results (Volume I): Teachers and school leaders as lifelong learners. OECD Publishing. https://doi.org/10.1787/1d0bc92a-en