AI in education isn’t some distant future; it’s already here and transforming the way that teachers teach. This includes one of the most important, but often invisible, aspects of teaching: scaffolding.
Scaffolding is the art of offering just enough support to help students access new concepts or skills, then gradually removing that support as confidence and understanding grow. It’s a core part of teaching and an area where AI can potentially make a meaningful difference.
So, how exactly can AI help teachers scaffold more effectively? Read on to explore:
- What the latest research reveals
- How to practically implement AI to help scaffold in your classroom
What does research say?
A recent study from researchers at Stanford University asked the question: Can AI help teachers scaffold middle school Maths curricula?
The researchers explored how Large Language Models (LLMs) such as ChatGPT could be used to design more accessible warmup activities or Do Nows – short starter activities that activate prior knowledge and set the tone for the lesson. Do Nows are a scaffolding tool that help bridge the gap between what students already know and what they’re about to learn.
The researchers started by conducting analyses of expert teachers, breaking down how they would scaffold lessons. What they found was that most undertook a three-part process:
- Observation – The teachers spotted areas in the curriculum that may be challenging for their students
- Strategy Formulation – They decided how to address those challenges (e.g., by reviewing prior knowledge or offering extra modelling)
- Implementation – They adapted or added materials to put the strategy into action
The researchers then tested how well LLMs could perform that third step, implementation, by generating warm-up Do Nows that were tailored to students who were performing below the expected standards.
Can AI help create scaffolding warm-up activities?
So, what did they find? When prompted effectively, the LLM-generated Do Nows were rated significantly higher by teachers, outperforming even those written by expert educators. Teachers rated them higher across four key criteria:
- Alignment to learning objectives
- Accessibility for students still developing foundational skills
- Readiness for immediate use in the classroom
- Overall preference
Practical tips for Scaffolding with AI in your classroom
The research shows real potential for LLMs to support curriculum scaffolding. So, how can you bring AI into your day-to-day teaching without feeling like you’re handing over your teaching to a robot? Here are three ways to get started:
1. Start experimenting
Begin testing out AI in your everyday tasks to see how it can (and can’t) improve your scaffolding and other tasks. This can help you get a sense of where it can streamline your workflow, where it falls flat, and how it might support the kinds of adjustments and scaffolds you already make for students.
Like any new tool, it takes some trial and error, but the only way to find the value is by getting hands-on. Check out our blog on approaches to using AI in the classroom for guidance on how to get familiar with it.
2. Prompt effectively
One of the biggest takeaways from the research was that the quality of the AI’s output depends heavily on the quality of the input. The most effective results in the study came when the LLM was given both the original curriculum materials as well as an expert-informed prompt to build from.
This relates to one of our 8 principles for effective AI use: developing your prompt craft. For example, simply asking, “Create a warm-up activity for a lesson on ratios,” is unlikely to yield good results. A more effective prompt might include the lesson objective, a description of students age and stage, and even the original warm-up task from the curriculum for reference.
3. Don’t skip the review
Even when AI offers helpful suggestions, your professional judgement is still front and centre. LLMs can produce biased, confusing, or occasionally flat-out wrong content, and that’s why refining is non-negotiable.
Think of it as a back-and-forth dialogue: you feed in your idea, ask the AI to build on it, then reshape what you get back. The best results don’t come from the first response, but from the iterative flow where you are bouncing ideas off one another as if it’s a co-intelligent partner.
So, before anything goes to students, take the time to review and revise. After all, if it’s going into your classroom, it should reflect your standards and voice.
Final thoughts on AI and Scaffolding
The idea of using AI to scaffold curriculum might sound futuristic or even a little unsettling. But in practice, it’s less about replacing teacher decision-making and more about speeding up the parts of the job that eat up your planning time. You’re still the expert; the AI tool simply helps you move faster.
