4 September 2026
From experimentation to evidence
Artificial intelligence has moved beyond the early adopter stage in schools. Many teachers already use AI to make help with planning, assessment and communication. Students increasingly use AI for homework, revision, coding and independent study. The challenge is no longer whether AI belongs in education. The challenge is deciding how to use it well.
That is why the Department for Education‘s decision to fund the development of national AI benchmarks for schools is significant. Educate Ventures Research has been awarded a £300,000 contract to create a digital and data skills pathway that will establish minimum and aspirational standards for the use of technology and AI across education.
The aim is not just to encourage more AI use. The aim is to help schools identify what effective, safe and educationally valuable use looks like. In a fast-moving market full of competing claims, that clarity is badly needed.
For teachers, this marks an important shift. The conversation is moving away from excitement about new tools and towards professional standards, evidence and accountability.
Workload relief remains the biggest opportunity
Few issues have shaped education in recent years as much as workload. Planning, assessment, administration and communication place constant pressure on teachers’ time. AI offers genuine opportunities to reduce some of that burden.
Many schools already use AI to:
- Draft lesson materials and worksheets.
- Produce differentiated resources.
- Generate reports and parent communications.
- Support formative assessment.
These applications often save time, particularly when dealing with repetitive tasks. However, the most effective use of AI is not replacing professional judgement but strengthening it.
A useful approach is to treat AI as a critical friend rather than a content generator. Instead of asking it to write an entire lesson, ask it to review part of a lesson you have already designed. Ask it to identify misconceptions that may arise. Ask it to suggest alternative explanations. Ask it to challenge assumptions in a sequence of learning.
This distinction matters. AI works best when it improves teacher thinking rather than replacing it.
The danger of the shortcut mindset
The promise of workload reduction can sometimes create unrealistic expectations. It is easy to assume that generating a lesson, presentation or worksheet in seconds automatically improves efficiency and quality. Experience suggests otherwise.
Generic AI-generated resources often resemble work produced by an inexperienced teacher. They may contain accurate information, but they rarely reflect the personality, expertise and classroom awareness that make teaching effective. They tend to lack the examples, explanations, diagrams, humour and carefully structured activities that experienced teachers develop over years of practice.
Students are becoming surprisingly adept at recognising AI-generated content. When a resource appears generic or formulaic, engagement can suffer. Learners often respond best when materials clearly reflect the expertise and effort of the teacher standing in front of them.
The most successful teachers therefore use AI to enhance their professional practice rather than outsource it. The technology can accelerate preparation, but it cannot replace the human element that turns information into learning.
AI as a learning partner
The government’s long-term vision includes AI tutoring systems that strengthen teaching and provide better information about pupil progress. At first glance, this can sound ambitious. Yet there are already examples of AI supporting learning in productive ways.
When designed carefully, AI tutors can provide instant explanations, offer additional practice questions, check understanding and help students overcome learning barriers at the moment they occur. A pupil working on programming at home no longer needs to wait until the next lesson to ask a question. They can seek clarification immediately and continue learning.
This is particularly valuable in Computing. Many concepts become difficult because misconceptions go unaddressed for too long. An AI system that provides timely feedback can help students stay engaged and maintain momentum.
Crucially, the most effective AI tutoring tools do not simply provide answers. They encourage thinking. They prompt students to explain their reasoning, justify decisions and refine their understanding.
The goal should never be faster answers. The goal should be deeper learning.
When easier learning becomes weaker learning
The greatest educational risk is not that AI gets answers wrong. It is that AI makes learning feel complete when understanding remains shallow.
Students can now receive an explanation, a summary, a piece of code or a completed paragraph within seconds. While this improves productivity, it can also remove much of the struggle that builds long-term understanding.
Learning depends on effort. Retrieval practice, problem-solving and productive challenge remain essential components of successful education. If students habitually rely on AI to perform tasks for them, they may develop the appearance of competence without developing the underlying knowledge and skills. They can produce work that looks impressive while remaining unable to explain, evaluate or apply what they have submitted.
This concern is particularly relevant in Computing, where mastering programming, algorithms and problem-solving depends on sustained practice. AI can make learning easier without making learners better.
Schools therefore need clear expectations around when AI supports learning and when independent thinking must take priority.
Accuracy still requires human judgement
Another common misconception is that AI can be trusted simply because it sounds confident. It cannot.
Generative AI systems continue to produce factual errors, misleading explanations and fabricated references. The most effective users of AI adopt a questioning mindset. They challenge responses, seek verification and ask the system to critique its own output. They treat AI as a knowledgeable but fallible assistant rather than an unquestionable authority. Students need to be taught this explicitly and teachers need to be mindful not to diminish their own art of reflection.
This approach mirrors good teaching practice. Critical thinking remains essential whether the source is a textbook, a website or an AI model. The responsibility for accuracy still sits firmly with the teacher.
Safety, safeguarding and data protection
As AI adoption increases, safeguarding and data protection become even more important.
Schools should be particularly cautious about entering pupil data into AI systems. Information relating to individual students, SEND needs or behavioural records require careful handling and appropriate safeguards.
The temptation to use free AI tools can create unnecessary risk. In many cases, free services generate revenue by collecting data or using user interactions to improve future models. For schools, the principle should be straightforward: understand where data goes, who has access to it and how it is used.
A strong AI policy should sit alongside existing safeguarding, behaviour and data protection policies. The government’s planned benchmarks will prove especially valuable here. Many schools do not need more AI products. They need greater confidence that the products they use are safe, appropriate and educationally sound.
Teachers remain at the centre
One message appears consistently across government policy, school guidance and classroom experience: teachers remain central to education. AI can provide feedback. AI can suggest explanations. AI can generate resources. AI can support tutoring. What it cannot do is replace professional judgement, relationships and expertise.
Great teaching depends on understanding learners, responding to misconceptions, building confidence and creating meaningful learning experiences. Those are fundamentally human skills. The future of AI in education is therefore unlikely to be about replacement. It is more likely to be about amplification.
The schools that benefit most will not be those that automate everything. They will be the schools that use AI to multiply the impact of effective teachers. The new national benchmarks offer an opportunity to move beyond the current Wild West of AI adoption and towards a more thoughtful approach, where innovation is balanced by evidence, safety and sound educational practice.
Want to know more? Watch our latest video on ‘At the chalk face’ where Dave is joined by Kat Morgan (our Lesson Hacker), an edtech and AI expert, to take a closer look at how AI is really being used in schools right now. From lesson planning and fact-checking to AI tutoring, assessment and personalised learning, they explore what AI can genuinely offer teachers and students – and where we need to be much more careful.
Looking for resources to help make the new academic year easier? Explore our Craig’n’Dave Resource Centre for ready-to-use Computer Science teaching resources, or use Smart Revise to support effective revision and reduce workload.
