15 Lesson 15 of 15
Nothing here breaks
This last lesson is the one the rest were for. You now know enough to experiment without being fooled, so the sensible thing to do is experiment.
I should probably learn this properly before I start. Surely there is a right way to use these tools, and some techniques to learn first. Until you have them, you will just be fumbling. Better to read up a bit more.
Those are two different dials. This whole site has been trying to turn one up and the other down.
See it in the lessonWhen 103 adults aged 50 and over were asked about learning AI, they were not reluctant. They thought it was important, they wanted to learn, and they knew why. What held them back was not knowing where to start. And they strongly preferred learning by doing over being lectured.
So the fumbling is the method. Nobody keeps a record of your clumsy attempts. Nothing you type damages the tool, costs anyone anything, or says anything about you. A chat that goes nowhere is closed and forgotten. This is an unusually safe thing to be bad at, and being bad at it for a while is how everyone who is now good at it got there.
What you have that you did not before
Confidence built on nothing is exactly what this site set out to avoid. What you have now is different, so it is worth spelling out. You can say where the words come from. You know one answer is a sample, and how to get a second. You can predict which tasks sit in the troughs. You know which store your words land in. You have four checks that cost almost nothing.
That is a working model. It lets you try something new and make sense of whatever comes back. That is the real skill, and it is why trying is now safe.
Ten things to try this week
- Paste in a long email and ask for the three things being asked of you.
- Give it a paragraph you wrote and ask for it two ways: shorter, and warmer.
- Ask it to explain something in your own field, and mark what it got wrong.
- Ask for twenty options for something, expecting to use one.
- Paste a recipe and ask it to scale for a different number of people, then check the arithmetic.
- Ask it to argue the opposite of something you believe, as well as it can.
- Give it a photo of a form or a letter and ask what it wants from you.
- Ask it to turn your rough notes into something you could send.
- Ask the same question in three fresh chats and compare.
- Ask it something you assumed was too odd to bother asking.
The last one matters more than it looks. The most common way to get little out of these tools is to only ask them things you already think they can do.
The one thing to keep
One finding behind this whole site is worth taking away on its own. In a survey of several hundred office workers, people with more confidence in the AI did less critical thinking. People with more confidence in themselves did more. Those are two different dials. This whole site has been trying to turn one up and the other down.
Those two dials are drawn below. Set them where you think you were before you started, then where you are now.
Read this diagram as text
Two sliders set confidence in yourself and trust in the tool, each from low to high. There are four regions. Low self-confidence with high trust: accepting whatever comes back. Low with low: not trying at all. High with high: trying a lot and checking little. High self-confidence with lower, measured trust: trying things and checking them, which is the goal. The line under the sliders names your region after every change.
Your week of trying things
The ten things from above, as a checklist. Ticks are kept in this browser only, so you can come back to it. Print it if you would rather have it on paper.
Ask it something you had assumed was too strange, too small, or too obvious to ask. That hesitation is the last thing between you and getting real use out of this.