AI exercises from the textbook: how to save a teacher hours

A teacher spends hours a week on two jobs that are not teaching: writing exercises and checking them. AI now does both well enough — if you set the right limits.

How it works

1. The source is the textbook, not the internet. The AI reads the very topic the class is on and writes problems at the same level, with the same terms. If the book is scanned, the AI reads the page images when the book is imported and extracts the topic's text from them.
2. Three kinds: multiple choice, numeric and short answer. The first two are checked mechanically, the third by the AI.
3. The teacher decides. During the lesson the teacher picks which exercise goes to the class. For homework the teacher picks the topic and the number of exercises, the system puts the set together, and the teacher can remove any exercise that doesn't fit.
4. After answering comes an explanation. The student can ask for a step-by-step explanation of the solution and ask a follow-up question. Before answering, the explanation is locked so the solution isn't given away early.

What stays with the teacher

• Explaining the topic and leading the lesson.
• The grade: the AI only marks an answer right or wrong; the teacher gives the grade.
• Deciding who gets which individual assignment.

What to be careful about

• AI makes mistakes. Rarely, but it does — especially in geometric figures and in the wording of word problems. In class, glance at an exercise before sending it; check homework right after you set it.
• Language. An exercise should be in the language of the textbook, not of the interface.
• Data. Ask the provider whether students' answers are used to train models.

The result

When exercises are written and checked automatically, every answer lands on the student's knowledge map. For the first time the teacher sees not the class average but who is behind on which topic.

In TutorVoice this is part of the lesson: see how it works or try it as a teacher.