Using AI to help students with difficult GCSE set books and exam comprehensions | Series 1, Episode 36

Episode 36 transcript

Using AI to help students with difficult GCSE set books and exam comprehension questions

AI is changing the way schools approach reading. Instead of silent text on a page, lessons can now sound more like conversations that pupils can follow. Teachers can feed a class text, a novel chapter or a news article into a tool such as Google NotebookLM and ask it to produce a script where two voices explain what’s happening. The dialogue can cover plot, tricky words, context and links to earlier lessons. Pupils listen first, then read and answer questions. It’s still the same material, simply introduced through talk. NotebookLM now creates podcast-style discussions from teacher materials and is open to under-18s in education settings.

AI can also build reading comprehension questions in the style of UK assessments. Teachers can ask for literal, inference and vocabulary items, complete with model answers, mark schemes and step-by-step reasoning. Pupils can query the AI about why an answer is wrong and get a clear, jargon-free explanation. That matches what the Education Endowment Foundation (EEF) says about digital tools that guide pupils as they practise reading strategies. The big gain is immediate feedback – something most classes miss because there just isn’t time.

For pupils who hesitate to ask for help, AI can be a quiet bridge. A child can speak or type into the system and hear or read a voice reply, without anyone else overhearing or looking at the first question. It’s a lifeline for those reading below class level, for EAL learners or for anyone who’s missed lessons. They can ask what a quote means, what a grammar term refers to or what a line implies. They can ask again if they need clarification. They can ask why and how. They can ask the AI to explain in a different way. They can keep asking until they get it, without fear of ridicule. Teachers and librarians can view the chat logs later and see where pupils are getting stuck. There are no stupid questions. There’s no fear of a dismissive human response. There’s no peer pressure, and, compared to a classroom setting, there’s all the time in the world.

Many pupils can read fluently but still struggle to turn that reading into structured written answers. Here AI can model the process. It can unpack the question, highlight task words, pick out what the examiner wants and suggest a short plan. It can flag useful quotes and show how to link them to points. This reflects the DfE’s guidance that AI can help with planning, feedback and resource creation, though staff must still check accuracy. Teachers can demonstrate the method with AI, then ask pupils to repeat the process themselves.

Personalised practice is another benefit. Studies on intelligent tutoring systems, including work from 2025 and earlier, show that adaptive AI tutors can improve reading and language scores when they adjust text difficulty in real time and offer regular, short sessions. A 2024 trial with 85 secondary pupils found that those using an AI reading platform achieved higher comprehension scores because the texts, glossaries and quizzes matched their level. UK evidence on EdTech tells a similar story: technology works when it provides practice with feedback, not when it simply adds more screen time. Schools can use this insight to schedule short daily reading bursts and group pupils based on AI reports. Ais such as ChatGPT and Gemini can keep a history of the chats. They can – on demand – outline learning gaps, concepts the student struggles with, and target areas to reinforce and revise.

AI can also design revision plans from pupil records. If a chatbot notices that a student repeatedly struggles with inference or keeps asking about metaphor, it can suggest a week of short, focused tasks on those skills. It can surface safe resources such as online explanations and videos, which teachers can review and edit before use. The result is a clear plan for homework or catch-up and, at the same time, evidence for teachers planning interventions.

Still, there are limits. AI can produce inaccurate information, present texts at the wrong level or tempt pupils to copy full answers instead of engaging with the source. DfE guidance reminds schools to check for accuracy, protect pupil data and make rules clear. Pupils need to know what’s AI-generated, what comes from the source and what must be in their own words. Learning to spot AI mistakes is now part of digital literacy.

Equity is another issue. The EEF warns that digital tools can widen gaps if only some pupils get guided use. Some AI reading tools work best in supervised or whole-class settings, not as unsupervised homework. Others lend themselves to individual study and use at home – but schools need to consider appropriate usage and safeguarding. Teachers need to consider how many pupils have access to reliable internet at home. Librarians and ICT staff can run drop-in sessions for those without devices or reliable broadband and can save safe, pre-checked prompts so pupils don’t have to start from scratch. That way, AI stays within the school’s safety net.

Teachers can also link AI work to long-term reading strategies. Start by teaching one of the EEF’s four: prediction, questioning, clarifying or summarising. Then ask pupils to use AI to generate five practice questions on that day’s chapter. After answering, they can ask the AI to explain any mistakes. It keeps the strategy alive every day without adding to marking and gives extra practice to pupils who need more steps to grasp each skill.

Academic integrity still matters. UK colleges and universities report that more than half of students now use AI for essays, and schools are seeing the same trend in Key Stage 4. Reading tasks should therefore encourage talking, annotation, short recorded reflections or in-class use only. Assessments can use unseen texts and timed reading so AI can’t complete them. Teachers, however, can still use AI behind the scenes to draft model answers and mark schemes. Whether it’s a teacher or AI showing pupils what a strong response looks like – and why – using language models as a free confidence building tutor remains one of the simplest, most effective uses of the technology.

Further reading and resources:

• Department for Education. Generative artificial intelligence (AI) in education. 12 August 2025.
• Department for Education. Generative AI in education: user research and technical report. 28 August 2024.
• Department for Education. AI in schools and colleges: what you need to know. 10 June 2025.
• Education Endowment Foundation. Reading comprehension strategies.
• Education Endowment Foundation. Using digital technology to improve learning.
• Létourneau, A. et al. A systematic review of AI-driven intelligent tutoring systems on K 12 students’ learning and performance. 2025.
• Hidayat, M.T. 2024. Effectiveness of AI based personalised reading platforms in improving reading comprehension among senior high school students.
• Google. 6 ways to use NotebookLM to master any subject. 8 September 2025.
• The Verge. Google is opening its NotebookLM AI tools to students under 18. 2025.
• Google’s NotebookLM AI homepage. For students, for teachers, and a slideshow presentation here.