Quick reference for the four numbers to measure, the four data traps, the cost of checking, the order effect, and three questions for deciding what to hand off.
🧭 Plot Summary
"AI saves time" is a claim, and you can test it. Take one task, do it yourself first, then do the same task with an AI, and write down what each way cost you. Most people count only the seconds the AI took to answer. They leave out the time spent writing the prompt, checking the answer, and fixing what was wrong.
Your mission is to measure all of it: minutes by hand, minutes with AI (prompt, wait, check and fix), a quality score against an answer key, and the total. A lunch survey with planted traps will show you what an AI does with messy data. A scheduling puzzle will show you a task where checking is cheap.
What you'll do in this activity
- Predict how long each way will take and which will score better, then find out.
- Answer five questions and draw a chart from a lunch survey by hand, with a timer, then do the same with AI.
- Hunt for four data traps planted in the survey and see what the AI did with them.
- Check the AI's answers against the data, and log the checking and fixing time as part of the cost.
- Try a scheduling puzzle where checking is cheap, and compare it with the survey, where it isn't.
- Learn three questions for deciding what is worth handing to AI, and what you should keep doing yourself.
Ground rules
- Do the by-hand version first, and put your pencil down before you open the AI.
- The data in this activity is invented. Keep real names, addresses and details about other people out of any AI tool.
- Record your times honestly. A rough number is fine, and the comparison is the point.
Why it matters
The Productivity Study later in this unit asks you to run three real tasks both ways and write up whether AI helped. This activity gives you the method: measure, check, and count the cost.
✅ Self-Check Before You Share
Check off each item as you get there. These aren't grades. They're your own signal.