An assistant describes plausible results before the class performs a test. I propose to distinguish prediction, simulation and observation.
In the example, the actual data is recorded separately and compared with the hypothesis. The produced text should not be presented as experience.
Technology can help formulate questions, but scientific documentation requires traces of the process. An unexpected outcome is information to be analyzed, not something to be corrected to make it look like the generated response.