Right AI™ Workshop Exercise
The Don't List
Your AI constitution needs hard lines, not vibes. Write prohibitions your AI can actually be trained on, then build your own list.
The Core Principle
A don't only trains your AI if a violation could be caught in a single output.
If you can't spot the violation by reading one response, neither can your AI.
The Standard · Three Tests
It names the exact behavior. Not a mood, not a value. A behavior.
You could spot a violation by reading one single output.
It says when and where the rule applies, and when it doesn't.
Your Turn · Build Your List
Five places don'ts live
Work one category at a time. Answer its question, write a don't, then run it through the tests. Aim for at least five across your list.
This check flags the patterns that usually break a don't. It is a prompt for your judgment, not a grade. Read each note, revise if you need to, then confirm the call yourself.
Confirm all three tests to add this one.
My Don't List
Your hard lines
Nothing here yet. Write your first don't above. Start with the category where you already know something has gone wrong.