The growing use of artificial intelligence in education is changing not only how Israeli students learn, but also how teachers assess them. A new study finds that many teachers are moving assignments back into the classroom, turning to written exams, face-to-face lessons and oral presentations rather than relying on homework and projects that students complete independently at home.
The study, published in the journal Teaching and Teacher Education, was conducted by Dr. Mital Amazleg and Prof. Gila Koru of the Faculty of Learning Technologies at Holon Institute of Technology, together with Professor Hadar and Prof. Yifat Ben-David Kolikant of the Seymour Fox School of Education at the Hebrew University of Jerusalem. Researchers interviewed 24 middle school teachers in Israel as part of the study.
Of the 24 teachers interviewed, 21 described situations in which they believed the use of generative AI tools should be limited or even completely prohibited. This included teachers who themselves regularly use AI in their teaching.
When teachers believe AI could undermine the purpose of an assignment, they said they prefer to move the task into the classroom and require students to demonstrate their knowledge and skills themselves. A central concern is the difficulty of determining who actually completed work submitted from home.
Teachers want grades to reflect what students know and have learned, rather than their ability to use AI tools to complete an assignment. The researchers describe this approach as "blocked pedagogy" - a situation in which teachers must determine for each assignment whether technology supports learning or allows students to bypass the skill they are supposed to develop.
One literature teacher who incorporates generative AI into some aspects of her teaching said she was considering replacing certain homework projects with classroom exams. She believed this would require students to demonstrate their own knowledge and understanding.
An English teacher in the study used AI even more extensively. She allowed students to practice spoken English through ChatGPT's voice mode and created chatbots that provide writing feedback tailored to each student's level.
However, when the objective was to develop independent writing skills, she restricted the use of AI. At the beginning of the school year, she also asked every student to write a text without assistance, establishing a baseline that could later be compared with their AI-assisted work.
The study indicates that concerns about AI are not necessarily the result of teachers being unfamiliar with the technology. One mathematics teacher with a master's degree in computer science specializing in artificial intelligence argued that generative AI could interfere with the development of mathematical thinking. In her view, the effort students put into reaching a solution is itself an essential part of the learning process.
The research also found considerable uncertainty among teaching teams. Fifteen of the 24 teachers described dilemmas about adapting education to the age of AI, while 17 said they felt that much of the responsibility for making decisions rested on their shoulders. Eleven said they felt they were dealing with the transformation largely on their own.
The researchers conclude that preparing teachers for the AI era must go beyond training them to use new technological tools. Teachers also need guidance in determining when AI enhances learning and when it effectively replaces the very skill an assignment is designed to develop.