Wissenschaftsunterricht mit KI

Science learning involves asking testable questions, designing investigations, measuring carefully, and interpreting evidence. AI can help students brainstorm hypotheses, suggest ways to visualize data, or identify possible sources of experimental error. That can support inquiry when the teacher anchors the task in observations students can verify. A chatbot can also suggest an untestable explanation, misread a data table, or turn a correlation into a cause. Start with the phenomenon, available materials, and learning goal. Ask students to record their own observations before consulting AI so they can compare its suggestions with what they saw. When using a generated hypothesis, require a measurable prediction and a plan for gathering evidence. For a data visualization, check that axes, units, sample size, and raw values are correct. Do not treat a smooth trend line as proof of a scientific explanation. AI can generate plausible experimental errors, but students need to connect each one to the actual setup. A suggestion about contaminated glassware is irrelevant if no glassware was used; a measurement error may matter if the class recorded temperature by hand. Have learners state why an explanation fits or does not fit their evidence, and compare results with trusted course materials or a knowledgeable instructor. Protect student data and follow school rules for any service. Avoid uploading identifiable student work or sensitive information without approval. Use AI as a discussion partner, not a hidden answer key. Assessment should make student reasoning visible through predictions, lab notes, diagrams, and explanations. Review whether AI use helps students ask better questions and interpret evidence, rather than simply producing more text.