Israeli Teachers Shift to In-Class Exams Amid AI Challenges in Education

A recent study reveals how artificial intelligence (AI) is transforming student assessment methods in Israeli schools. Conducted by researchers from Holon Institute of Technology and the Hebrew University of Jerusalem, the study highlights that about 90% of teachers limit or prohibit students' use of AI tools like GenAI during assignments. This shift aims to ensure that grades accurately reflect students' own knowledge and skills, as AI use raises concerns about who actually completes the work.

The research, published in "Teaching and Teacher Education," involved semi-structured interviews with 24 middle school teachers across Israel. Most teachers reported a preference for traditional in-class written exams, oral presentations, and frontal classroom activities over take-home projects, which are more vulnerable to AI assistance. This approach, termed "pedagogy of blocking" by the researchers, involves evaluating whether AI use advances learning or merely allows students to bypass essential skill development.

Some teachers who integrate AI into their teaching still restrict its use when independent demonstration of knowledge is required. For example, an English teacher used AI tools for spoken language practice but required initial writing assignments without AI help to establish baseline skills. Another literature teacher considered replacing some home projects with in-class tests to ensure authentic student performance.

The study also found that resistance to AI is not due to lack of technological familiarity. A math teacher with a master's degree in computer science specializing in AI expressed concern that AI use might hinder the development of mathematical thinking, emphasizing the importance of students' effort in problem-solving.

Additionally, many teachers expressed uncertainty about adapting teaching methods to the AI era. Fifteen out of 24 mentioned dilemmas in adjusting instruction, 17 felt the responsibility for decision-making largely rested on them, and 11 described feeling isolated in managing these changes. The researchers stress that teacher training must go beyond learning new tools to include guidance on when AI supports learning and when it undermines skill acquisition.