AI tools can produce a lesson plan in about 30 seconds. For a primary teacher working through a planning session on a Tuesday evening, that is an appealing proposition. But the speed at which these tools generate content is also part of the problem: they produce plausible-sounding material quickly, and plausible is not the same as accurate.
This is not an argument against using AI for planning. Used well, these tools can save time on the structural and administrative side of lesson preparation. But in science, where a misconception embedded in a lesson can take months to undo, there are specific things worth checking before any AI-generated content goes near a class.
1. Is the science actually correct?
This is the most important check, and the one most likely to be skipped when time is short.
AI language models are trained to produce text that reads well and sounds authoritative. They are not trained to be factually accurate, and they have no awareness of when they are wrong. In primary science, this produces a specific kind of error: confident, well-presented content that contains a scientific inaccuracy the teacher may not catch, and that pupils certainly will not.
Common problem areas include:
- Explanations of forces, energy, and electricity, where everyday language and scientific language diverge significantly
- Life cycles and classification, where AI tends to oversimplify in ways that create misconceptions
- Explanations of physical processes such as evaporation, condensation, and dissolving, where the correct scientific explanation is counterintuitive
If your own subject knowledge in the relevant area is secure, checking an AI-generated explanation takes two minutes. If it is not, check against a reliable source before using the content: a good primary science scheme, the ASE’s resources, or a subject-specific reference you trust. Check the lesson against PLAN assessment’s knowledge matrices.
2. Will the activity actually work in a primary classroom?
AI tools do not know your classroom. They do not know what equipment you have, how your class responds to group work, how much time you actually have in a given lesson, or whether the practical activity they have suggested requires resources that do not exist in most primary schools.
Read any suggested practical activity as a practitioner, not just as a subject check. Ask:
- Do we actually have this equipment, in sufficient quantities?
- Is the level of independence this assumes realistic for my class?
- Will this work in the time available, including setup and clearing away?
- Are there any safety considerations the AI has not flagged?
AI-generated activities tend to assume ideal conditions. Classrooms rarely provide them. A quick read-through from a practical standpoint will usually surface any issues.
3. Does it actually cover the right National Curriculum content?
AI tools can drift. Ask for a lesson on sound for Year 4 and you may get something that covers the topic broadly but misses the specific aspects required by the National Curriculum programme of study: that sounds are made by vibrations, that vibrations travel through a medium, that pitch and volume can be changed. The lesson looks like a sound lesson. It may not be teaching the right things.
This matters more in science than in some other subjects, because the NC programme of study for science is quite specific about what should be taught in each year group. A lesson that addresses the general topic without covering the required content will leave gaps that are difficult to spot and harder to fill later.
Keep the relevant section of the NC open alongside any AI-generated plan and check that what the lesson actually teaches matches what the curriculum requires. This takes less time than it sounds.
Checklist
A quick checklist: before using an AI-generated science lesson,
- check the science is factually accurate;
- check the practical activity is workable in your classroom;
- check the content matches the National Curriculum programme of study for the relevant year group.
- also – always remember to do a risk assessment, and even better if you can check any activities against CLEAPSS guidance for safety.
If all these pass, the lesson is probably worth using as a starting point.
The bigger picture
None of this means AI planning tools are not useful. For generating a lesson structure, producing differentiated questions, or drafting a sequence of activities as a starting point, they can genuinely save time. The problem is not the tool; it’s using it without the professional judgement that turns a generated draft into a lesson that actually works.
In primary science in particular, that judgement matters. The subject is one where teacher confidence has a direct effect on how much gets taught and how well. Using AI to bypass subject knowledge rather than support it is likely to make that problem worse, not better.
Further reading
For more on AI in primary science, check out this article.
If you’re finding the lessons you get are not very good, then you might want to make your prompt better. This guide explains more.
For more AI guides, check out my YouTube channel
The post AI-generated science lessons: 3 things to check appeared first on Danny Nics Science Fix. Written by Danny Nicholson