Using AI Responsibly
AI is part of how we investigate and think. It is a tool, not the deliverable.
The principle
Use AI to brainstorm causes and options, summarize the history you've gathered, research typical costs, and draft recommendations or approval packages. Then read it, verify it, and own it.
You own the output
If you ship something AI produced, you are accountable for it exactly as if you wrote it yourself.
Think of Tesla self-driving: if the car is driving and you get into an accident, it's still your fault — you can't blame the autopilot. Same here. An AI result is not an excuse. You must be able to answer every question about it and justify it as your own work.
Good uses
- Brainstorming possible causes and resolution options.
- Summarizing the GL entries, work-order history, or email thread you pulled.
- Researching typical cost, lifespan, and failure modes.
- Drafting a first version of an approval package or owner email.
Not acceptable
- Pasting a prompt result and shipping it unread.
- Citing "the AI said so" as the reason for a decision.
- Trusting AI figures, citations, or lease interpretations without verifying them against the source.
Verify before you ship
- I read the entire output.
- I checked every number, date, and citation against the real source (GL, lease, quote).
- I confirmed it fits this specific property, lease, and situation.
- I can explain and defend every part of it.