Dr Mandy Pierlejewski
In this blog post, I explore the idea that using generative AI to plan lessons is cheating. I compare artificial intelligence lesson planning with the use of government validated schemes such as Little Wandle, using Foucualt’s work on discipline to explore notions of lesson authorship. I suggest that the use of generative AI to plan lessons disrupts disciplinary regulation and can therefore be seen as a form of resistance.
I visited a school recently as part of my role as a teacher educator. I had suggested to the student that they could use generative AI to help them plan lessons as they were spending many hours producing university lesson plans. A conversation with teachers at the school led to a discussion of whether this was a kind of cheating as the lesson would not be the students’ own work. I found this very interesting as almost all primary schools in England follow the government validated schemes for teaching systematic synthetic phonics (SSP) (Department for Education, 2023) and yet this planning, which is not generated by the teachers themselves, is not seen as “cheating” or “not the students’ own work”.
Following a scheme for the teaching of phonics, maths or any other subject is increasingly required in schools. Anecdotal evidence indicates that big multi academy trusts, in particular, seem to favour this approach in an attempt to ensure consistency of delivery across the trust. In such cases, the teacher delivers the content of the scheme, using a predetermined lesson plan. Some schemes and schools focus specifically on fidelity to the scheme, giving little autonomy to the teacher to adapt their teaching or use the plan creatively. This, however, is not seen as cheating.
Using generative AI to plan lessons is gradually becoming a part of some teachers’ planning approach but, in my experience, is not yet widely used. When teaching initial teacher education students to plan lessons, I explain that AI can be like a teaching assistant which can work with the student to co-plan a lesson. The students need to be critical in evaluating the lesson plans generated by AI and adapt them to ensure they work for them. This approach to supporting student teachers to plan using AI was also reported in van den Berg and Plessis’ (2023) research. I have also planned with AI myself, and have found that the dialogue between human and artificial intelligence produces innovative ideas. I compare it to planning with an experienced and creative colleague. Anecdotal evidence, however, would indicate that this is not how it is seen in schools. Why might this be?
To explore this question, I have used Foucault’s (1977) work on disciplinary power. Using a Foucauldian lens, it appears that the use of generative AI to plan lessons disrupts the disciplinary regime which regulates individuals in schools. Foucault, in Discipline and Punish (1977), identifies three aspects of disciplinary power: hierarchical observation, normalising judgements and the examination.
Hierarchical observation
Teachers are regulated through hierarchical observation techniques such as lesson observations, lesson feedback, book scrutinies, data analysis and planning. When using a scheme such as the government validated scheme “Little Wandle” (Little Wandle Learning Trust, 2024), this surveillance takes the form of observations which focus on fidelity to the scheme. Teachers are judged on how closely they have adhered to the lesson outlined in the Little Wandle plan. Even though the teacher has not planned the lesson, surveillance can take place because what is being measured is the fidelity to the scheme.
A lesson planned with generative AI is much more difficult to evaluate. A scrutiny of the teacher’s lesson plan would not reveal how much of the lesson had been planned by the teacher and how much by the AI. This thwarts the hierarchical observation process as it is unclear to the observer whether it is the AI which is under surveillance or the teacher. This therefore, could lead to the perception of cheating as there is always the possibility that the lesson is good because the AI generated content is good. The judgement system fails when the object if surveillance is blurred. Is it the human? Is the AI? Or is it a cyborg fusion of the two?
Normalising judgements
The disciplinary system in schools relies on comparing teachers to norms. Traditionally, these have been embodied in the Teachers’ Standards (TS) (Department for Education, 2013) and hierarchical observation focused on how closely teaching adhered to these norms. The use of prescriptive schemes has required new norms to be devised as these lessons cannot be compared to TS4 “plan and teach well structured lessons” (p11) as the plan is not created by the teacher. In the current teaching of SSP and many other subjects, the new norm is fidelity to the scheme. Rather than evaluating the teacher’s plan, observers evaluate the level of fidelity to the scheme. Thus, teaching becomes a process of delivering schemes rather than planning and delivering well structured lessons.
Using generative AI to plan lessons disrupts the normalisation process. Lessons cannot be evaluated against TS4 “Plan and teach well structured lessons” because it is unclear who has planned the lesson. The quality engendered in a lesson plan cannot be accurately attributed to the teacher and therefore, it is unclear whether it is the teacher or the AI who has met the norm. New norms relating to the collaborative process of co-production of lesson planning with AI have not yet emerged. This therefore, leads some to view AI lesson planning as cheating as it cannot be compared to norms which determine the quality of teaching.
The examination
The combination of hierarchical observation and normalising judgements is the examination (Foucault, 1977)- the performance of knowledge which demonstrates the quality of teaching. In terms of education, this is embodied in the formal lesson observation. Lesson observations which focus on the teaching of SSP replace judgements about planning with judgements about fidelity. The teacher performs the lesson just as the scheme determines. The closer it is to the scheme, the better the lesson. The judgment of quality using this combination of hierarchical observation and normalising judgements is simple- if the lesson accurately replicates the scheme, it is a good lesson.
The performance of a lesson planned with AI is much more problematic to evaluate. It is unclear to the observer how much if the performance is the teacher and how much is the AI. Rather than one actor on the stage, there are two- the teacher and the AI. Is the teacher like a puppet, being controlled by the AI and delivering an AI generated lesson which they claim to be their own? Is the observer observing some kind of collaborative dance between the AI and the teacher? Is the performance hiding the AI in the shadows?
Resistance
Such questions remain unresolved as the process of planning with AI cannot be made visible. Discipline requires the invisible to be made visible so that it can be judged (Foucault, 1977). All aspects of teaching must come under the gaze of the school and beyond for judgements about quality to be made. AI planning evades this trap of visibility and it could therefore be seen as an act of resistance as it challenges and disrupts the whole process of discipline which exists in schools.
References
Department for Education (2013) Teachers’ Standards: Guidance for School Leaders, School Staff and Governing Bodies. London: Crown. DOI: DFE-00066-2011.
Department for Education (2023) Choosing a phonics teaching programme [Online] Available from. <https://www.gov.uk/government/publications/choosing-a-phonics-teaching-programme/list-of-phonics-teaching-programmes> [Accessed on 2nd March 2024]
Foucault M (1977) Discipline and Punish. Harmondsworth: Penguin Books Ltd.
Little Wandle Learning Trust (2024) Teach reading: change lives! [online] Available from: <https://www.littlewandlelettersandsounds.org.uk/> [accessed on 2nd March 2024]
van den Berg G and du Plessis E (2023) ChatGPT and Generative AI: Possibilities for Its Contribution to Lesson Planning, Critical Thinking and Openness in Teacher Education. Education Sciences 13(10). DOI: 10.3390/educsci13100998.
Altered States of Academia 
Leave a comment