Science: Experiments, Explanations, Quizzes

Three core needs in experimental science

Science subjects are a natural fit for ChatGPT because they combine three specific needs the tool handles well: designing experiments within tight material constraints, explaining abstract phenomena to widely varying age groups, and checking understanding regularly without eating up your evenings. The first two support your lesson planning at every level. The third can involve students directly, following the framework introduced earlier in the course: in class from grade 8 (4e) onward, and with supervised independence in high school.

Design a protocol with the equipment you have

Start with the biggest time-saver: designing protocols that match the equipment actually in your storage closet and your students’ level. The key is a single prompt that lists every constraint upfront, leaving nothing for the model to guess.

I am a high school physics and chemistry teacher.
I want to run an experiment on light refraction.
Available equipment: light source, plexiglass half-cylinder, protractor, graph paper.
Propose a protocol in 6 steps maximum, including safety instructions.
Students are 15-16 years old; 45 minutes for the hands-on part.

Notice the structure: who you are and which class you teach, the scientific objective, material constraints, the expected format with safety included, and the time limit. Remove any element, and the model fills the gap with an assumption—often a bad one. You might end up with a twelve-step protocol for a forty-five-minute session.

This approach also helps when ideal equipment is missing. Describe what you have, and ChatGPT will suggest a substitute experiment, a complementary simulation, or a video demonstration for anything you cannot do in class. Remember, a model suggestion is not a validated protocol. Always cross-check its safety instructions against your lab’s official rules, which take precedence.

The right analogy, at the right age

The second use case is explanation. ChatGPT can tailor an analogy to your audience’s age. Take a concept that spans the curriculum, such as plant nutrition, and ask for three explanations in a single prompt: one for elementary, one for middle school, and one for high school. Specify what each level has already studied. You get material to build a learning progression or differentiate within a mixed-ability class.

Review the output with a didactic eye. A convenient analogy can cement a misconception that the next level will have to unlearn. The classic example is the “mini solar system” model of the atom: easy to remember, but wrong about the nature of electrons, which do not follow orbits. Ask the model to list the limits of each analogy, and choose the one that prepares students for what comes next, not just the one that sounds clever today.

Explanation also happens through diagrams. ChatGPT can describe diagrams for you or your students to reproduce: the step-by-step construction of an electric circuit, or the labels for a plant cell drawing. Having students draw from a description is itself a powerful comprehension check. If a student places an element incorrectly, they misunderstood the text.

Multi-level quizzes in the same hour

Finally, assessment. ChatGPT generates quizzes of increasing difficulty on the same topic, making differentiation simple. Ask for ten questions on photosynthesis in three versions: a recognition level with clear-cut multiple-choice answers; a comprehension level with short open-ended questions; and an application level requiring reasoning. While advanced students tackle the third level, others consolidate their knowledge with the first.

For an end-of-chapter quiz, provide your chapter outline or the relevant curriculum excerpt. ChatGPT will then cover all learning objectives instead of picking randomly from the theme. Always review every question alongside its answer key. A single error in a multiple-choice question propagates to the whole class.

In the field: Life and Earth Sciences, Physics and Chemistry

In Life and Earth Sciences (SVT), the most fruitful uses revolve around classifying living organisms, where students describe an organism and compare the response to the dichotomous key in their textbook; genetics, with simulated crosses; and ecology, through food chains and the impact of disturbances. In geology, explain plate movements and their consequences, or trace the water cycle by following a single drop through a pollution scenario.

In physics and chemistry, keep three reflexes in mind. For any calculation, select a higher reasoning level in the model selector—Medium or High on the Thinking slider, or the Think button on Free and Go plans—and ask the model to show its steps. Before any hands-on work, compare the suggested instructions to your lab’s official rules. To bring the subject to life, ask for the historical context of a discovery, but verify names and dates against a reliable source. This is where plausible errors slip in most easily.

Your turn

Build a kit for one chapter.

1. Choose a chapter from your curriculum.
2. Have the model design a protocol using your actual equipment, then validate it against your lab rules.
3. Request three explanations of a key concept for three different levels, including the limits of each analogy.
4. Generate a ten-question quiz with three difficulty levels.
5. Review everything line by line. In science, a plausible, well-formulated error is the worst trap.

Official sources: Getting started with ChatGPT as an educator · GPT-6 and other models in ChatGPT (OpenAI Help Center) · Artificial intelligence and its uses in education (éduscol)

Key takeaways

• ChatGPT helps design protocols adapted to available equipment, which you must validate against lab rules.
• Choose an analogy based on the student level and how well it prepares them for the next stage.
• For calculations, increase the reasoning level and require explicit steps.
• Three-level quizzes enable differentiation within a single class period.
• Your subject-matter expertise remains the final filter, and it cannot be delegated.