Seven weeks ago, Maya, Ben and Rosa each asked the same thing. What is this tool, and can I trust it?
Now they know where it goes wrong, what never to paste, how to check and how to stay honest.
This week each of them writes a one-page AI plan. Maya's is for her family, Ben's for his class and Rosa's for her shop.
Every plan in this course answers the same four questions.
What will we use AI for? What does it never see? Who checks its work? How do we tell people?
One page is enough. A short plan that people follow beats a long one nobody reads.
| Question | Where the course covered it |
|---|---|
| What will we use AI for? | Week 1 (what AI is) and week 4 (ask well) |
| What does it never see? | Week 3 (the never-paste list) |
| Who checks its work? | Weeks 2, 4 and 5 (errors, checking, fairness) |
| How do we tell people? | Week 7 (honest use) |
| What about scams? | Week 6 (code word and call back) |
The SBA gives small businesses simple advice that fits any plan. AI is relatively new, so start small.
It also suggests testing tools to see if they add value. Pick one or two tasks, not everything at once.
NIST lists what makes an AI system trustworthy. It should be valid and reliable, safe, secure and resilient.
It should be accountable and transparent, explainable and interpretable, privacy-enhanced, and fair with harmful bias managed.
You cannot test a tool for all of these at home. But your plan can ask the same things of how you use it.
| NIST trait | What your plan can ask |
|---|---|
| Valid and reliable | Do we check facts before we rely on an answer? |
| Privacy-enhanced | Do we keep private details out of the tool? |
| Accountable and transparent | Do we say who checked it and that AI helped? |
| Fair, with harmful bias managed | Do we look for who might be missed or misjudged? |
| Statement | True or false? |
|---|---|
| A one-page plan can be enough. | ? |
| The SBA says to start with as many AI tasks as possible. | ? |
| NIST lists fairness, with harmful bias managed, as a trustworthy trait. | ? |
| A good plan says who checks AI work. | ? |
Sources: US Small Business Administration, AI for small businesses · NIST AI 100-1, AI Risk Management Framework (2023)
Tomorrow: the most important line in any plan, who checks.