Inclusive education in the age of AI: Adaptive teaching strategies using ChatGPT

ANNA CHATWIN, DEPUTY HEAD OF TEACHING, LEARNING & INNOVATION AND TEACHER OF HISTORY & POLITICS, BERKHAMSTED SCHOOL, UK

The rapid advancement of generative AI presents educators with a pressing question: how can artificial intelligence (AI) be harnessed to foster inclusive and adaptive teaching practices? Research by Mollick and Mollick (2023) provides a tangible framework for upskilling students in the use of large language models (LLMs), enabling personalised learning experiences tailored to meet specific needs.

The approach

A small cohort of 11 sixth-form students were successful in their applications to participate in Berkhamsted School’s first bespoke pilot AI course in February 2024, designed to help them to learn both with and about AI. The goal was to enrich their learning by developing ‘augmented intelligence’, exploring the synergies between human and artificial intelligence to maximise their potential.

The course facilitated experimentation with the seven different learning modes explored by Mollick and Mollick (2023): AI-Tutor, AI-Mentor, AI-Simulator, AI-Student, AI-Teammate, AI-Coach and AI-Tool. This reflection piece evaluates the efficacy of the three modes rated most effective by the students as adaptive teaching strategies.

AI-Tutor

Bloom (1984) stated that, despite the clear performance gap, one-to-one tutoring was ‘too costly for most societies to bear on a large scale’ (p. 4). However, the asynchronous use of an AI-Tutor affords students the opportunity to engage with personalised instruction and immediate feedback, something unattainable for a single teacher to provide alone. 

Below is a Key Stage 3 history prompt that was easily adapted by sixth-form students to facilitate a tutoring experience of an A-level standard: 

I am a Year 8 history student learning about the transatlantic slave trade who has studied different abolitionists, including […]. You are an upbeat and encouraging tutor who adopts a Socratic approach of only asking questions. Ask me questions about the different contributions of these activists, including their methods, achievements and historical significance. If I answer a question accurately, praise me. If I am struggling, be encouraging and give hints to help me. When pushing me for information, close your responses with a question so I keep generating ideas. Rules: assess my prior learning about the topic and adapt your questions to my level of understanding.

AI-Student

By employing an AI-Student, learners benefit from the protégé effect, reinforcing their knowledge through explanation, elaboration and correction. 

This prompt was crafted for Key Stage 5 politics, but it is easily transferable across secondary subjects: 

You are an A-level politics student who is struggling to grasp the complexity of the Electoral College and has many questions. Your teacher is ready to help to clarify any confusion you have. Engage in a role-play where you as the student ask questions about the technicalities of the Electoral College, while your teacher provides explanations and guidance to enhance your understanding. Remember, you can only ask questions as the student, and your goal is to deepen your comprehension of this topic. In each interaction, you can indicate the extent to which the teacher’s responses are helping to aid your understanding.

AI-Simulator

Students found this to be the most abstract mode, creating bespoke AI-Simulators to enable immersive, scenario-based practice. This enhances critical thinking and bridges the gap between recall of technical knowledge and practical application skills. 

The following example was designed for Key Stage 4 history but has strong potential across other disciplines:

I am a Year 10 pupil who is learning about the Cuban Missile Crisis as part of my studies of the Cold War. My task is to imagine that I am an advisor to President JF Kennedy and come up with a response to the placement of Soviet missiles in Cuba. You should play the role of President Kennedy. The aim of this interaction is to allow me to practise hypothetical military planning, while considering the following non-negotiable geo-political factors: […]

The impact

Each week, one mode was presented, modelled and subsequently trialled by the students, with feedback collected via surveys and focus group discussions at the end of the course. The AI-Tutor mode was most popular, particularly valued for its ability to provide targeted one-to-one support. Several students noted the effectiveness of the AI-Student in revealing weaknesses through demanding externalisation of knowledge. This often led students to recognise and address gaps that they hadn’t realised existed, a common symptom of the cognitive bias known as the Dunning-Kruger effect, which distorts metacognition (Kruger and Dunning, 1999).

The AI-Simulator required most creative thinking and was rated the least popular of the three. However, one physics student highlighted its unique strength in enabling problem-solving, including step-by-step workings, as a way of deliberately practising for upcoming exams.

Despite acknowledging certain limitations in the use of LLMs, including the risk of confabulation or hallucination, all students indicated a commitment to using different modes explored as part of their subsequent A-level study. In this context, confabulation refers to the generation of plausible yet fabricated information, while hallucination describes inaccurate information resulting from insufficient training data or inherent biases in the data used.

Reflections

Facilitating this experimentation in my school context has demonstrated that transferring these approaches across key stages and subjects is simple, impactful and efficient, due to the scalability and adaptability of these prompts. Encouraging the development of ‘augmented intelligence’ has allowed every student to engage meaningfully in the learning process, while maintaining the teacher as the essential ‘human in the loop’.

However, challenges remain in ensuring that students do not become overly reliant on AI and that they critically engage with its outputs. AI is of limited use to beginners who are entirely new to a subject. According to the novice-expert phenomenon, novices lack the foundational schemas and domain-specific knowledge needed to critically evaluate AI-generated content or even to prompt it effectively. Without a solid grounding in the subject, they may struggle to distinguish between accurate and misleading information, reinforcing misconceptions rather than deepening understanding. This highlights the necessity of expert guidance in the learning process, ensuring that AI serves as a tool for enhancement rather than a crutch for superficial engagement.

The explicit and intentional upskilling of students’ use of AI chatbots has led me to reflect on the broader implications of effective and thoughtful AI integration in secondary education. As educators, we have a responsibility to guide AI’s role in education, ensuring that it enhances learning in ways that are inclusive, adaptive and transformative.

The examples of AI use and specific tools in this article are for context only. They do not imply endorsement or recommendation of any particular tool or approach by the Department for EducationThe ministerial department responsible for children’s services and education in England or the Chartered College of Teaching and any views stated are those of the individual. Any use of AI also needs to be carefully planned, and what is appropriate in one setting may not be elsewhere. You should always follow the DfE’s Generative AI In Education policy position and product safety expectations in addition to aligning any AI use with the DfE’s latest Keeping Children Safe in Education guidance. You can also find teacher and leader toolkits on gov.uk .