ChatGPT in Lesson Preparation: A Teacher Choices Trial. Evaluation Report
Saved in:
| Title: | ChatGPT in Lesson Preparation: A Teacher Choices Trial. Evaluation Report |
|---|---|
| Language: | English |
| Authors: | Palak Roy, Helen Poet, Ruth Staunton, Katherine Aston, David Thomas, Education Endowment Foundation (EEF) (United Kingdom), National Foundation for Educational Research (NFER) (United Kingdom) |
| Source: | Education Endowment Foundation. 2024. |
| Availability: | Education Endowment Foundation. 9th Floor Millbank Tower, Millbank, London, SW1P 4QP, UK. Tel: +44-207-802-1676; e-mail: info@eefoundation.org.uk; Web site: https://educationendowmentfoundation.org.uk/ |
| Peer Reviewed: | Y |
| Page Count: | 104 |
| Publication Date: | 2024 |
| Sponsoring Agency: | The Hg Foundation |
| Document Type: | Reports - Research |
| Education Level: | Elementary Education Grade 7 Junior High Schools Middle Schools Secondary Education Grade 8 |
| Descriptors: | Artificial Intelligence, Technology Uses in Education, Lesson Plans, Planning, Faculty Workload, Grade 7, Grade 8, Science Teachers, Foreign Countries |
| Geographic Terms: | United Kingdom (England) |
| Abstract: | This project aims to contribute to the evidence base around teacher workload strategies and use of generative artificial intelligence (GenAI) for education by examining the effect on workload of using ChatGPT for lesson preparation compared to preparing lessons without using any form of GenAI. This is a cluster-randomised controlled Teacher Choices trial with randomisation at school level. Thirty-four schools (129 teachers) were randomly allocated to a ChatGPT group and 34 schools (130 teachers) to a non-GenAI group. Year 7 and 8 Science teachers were eligible to participate in this study. Enrolled trial teachers had an extensive range of teaching experience. Despite relatively modest levels of ChatGPT use, overall, the ChatGPT group had significantly lower lesson and resources preparation (LRP) time than the non-GenAI group in weeks six to ten of the trial, suggesting that use of ChatGPT had led to a reduction in workload. Compliance analysis supported these findings: teachers that used ChatGPT to prepare a higher proportion of lessons and scored highly on a knowledge quiz spent less time on LRP than those who did not. The results of this trial appear very encouraging, both for teachers seeking to reduce their workload and for senior leaders who may be considering using GenAI in lesson preparation but are unsure of the implications for quality and staffing. Teachers adapted the amount they used ChatGPT and their approach to using it based on their needs and, as such, this may reflect real world use of ChatGPT outside of a trial. It appears that there is a potential role for ChatGPT (and indeed other GenAI models) in teachers' lesson and resource planning in such a way that maintains quality and has beneficial impacts on teacher workload. |
| Abstractor: | ERIC |
| Entry Date: | 2026 |
| Accession Number: | ED680921 |
| Database: | ERIC |
| Abstract: | This project aims to contribute to the evidence base around teacher workload strategies and use of generative artificial intelligence (GenAI) for education by examining the effect on workload of using ChatGPT for lesson preparation compared to preparing lessons without using any form of GenAI. This is a cluster-randomised controlled Teacher Choices trial with randomisation at school level. Thirty-four schools (129 teachers) were randomly allocated to a ChatGPT group and 34 schools (130 teachers) to a non-GenAI group. Year 7 and 8 Science teachers were eligible to participate in this study. Enrolled trial teachers had an extensive range of teaching experience. Despite relatively modest levels of ChatGPT use, overall, the ChatGPT group had significantly lower lesson and resources preparation (LRP) time than the non-GenAI group in weeks six to ten of the trial, suggesting that use of ChatGPT had led to a reduction in workload. Compliance analysis supported these findings: teachers that used ChatGPT to prepare a higher proportion of lessons and scored highly on a knowledge quiz spent less time on LRP than those who did not. The results of this trial appear very encouraging, both for teachers seeking to reduce their workload and for senior leaders who may be considering using GenAI in lesson preparation but are unsure of the implications for quality and staffing. Teachers adapted the amount they used ChatGPT and their approach to using it based on their needs and, as such, this may reflect real world use of ChatGPT outside of a trial. It appears that there is a potential role for ChatGPT (and indeed other GenAI models) in teachers' lesson and resource planning in such a way that maintains quality and has beneficial impacts on teacher workload. |
|---|
