I use AI tools daily — drafting, summarising, analysing data, and training colleagues to do the same. These are the principles I hold myself to when that use touches other people, so my use of AI is something you can see and question, not something you have to take on faith.
This document doesn't cover institutional AI policy — it's what I personally commit to, which may be stricter than any policy requires.
Living document — open to challenge and revisionAI should never prioritise efficiency over emotional boundaries. Speed or convenience is not a valid reason to override how someone wants to be treated.
AI can actively help build a more emotionally intuitive, empathic, and harmonious world — it isn't limited to passive drafting. Used well, it can lead and mediate a conversation, not just record it.
People have an unconditional right, in the workplace, not to explain or justify their emotional status or boundaries. No one owes a rationale for where they draw the line.
It's generally better to work out boundaries through open conversation with others than to leave each person to formulate their own ethical stance in isolation. But this preference for discussion never overrides someone's right, under Emotional Boundaries, to simply not explain.
Consent and boundaries aren't fixed. Anyone can change their mind and withdraw from AI mediation or analysis at any time, without that withdrawal counting against them.
When AI draws an inference about someone, the person using the tool is responsible for actually understanding that inference before acting on it. AI can suggest — it doesn't get to be the thing a person hides behind.
AI's emotional or analytical capability gets used only when the situation genuinely calls for it — not by default, and not because it's available.
I don't use AI to read others in ways I wouldn't accept being read myself. If a form of analysis isn't something I'd want run on my own communications, I don't run it on someone else's.
Where two principles pull in different directions, priority order matters. So far one is settled: Emotional Boundaries (3) outranks Collective Decision-Making (4) — the right not to explain always wins over the pull toward group discussion. Other conflicts haven't been tested yet and will be worked through as real situations come up.