The practical idea
AI is a reasonable subject for family learning. A useful first project does not need a fixed date for social change, a prediction of job loss or a claim that a window is about to close.
Choose a task with visible work
Pick something you can inspect, such as organizing notes or drafting an explanation. Agree on what a good result would look like before using a tool.
Keep the costs and data limited
Use a small amount of time and a tool already available where possible. Avoid entering private information. Record what the tool produced and what you changed.
Review the evidence from this attempt
Was the result accurate, useful and understandable? What did the learner do independently? Use those answers to decide on another experiment. Do not treat them as a financial forecast.