Using AI responsibly and inclusively at work

Most people aren’t reckless with AI and they aren’t refuseniks either — they’re somewhere in the middle, using it day to day and quietly wondering whether they’re doing it well. That’s the honest place to start. The question isn’t whether you use AI; it’s how. And the answer comes down to a handful of habits you can build into ordinary work, no technical wizardry required.

These are the six places responsible, inclusive AI use shows up most — the same six the free “Do you use AI responsibly and inclusively?” self-check measures. Read them as a set of practical instincts, not a compliance checklist.

1. Stay alert to bias

AI learns from the world as it is, which means it can encode and scale the very biases we’re trying to undo. A tool can be confidently, fluently wrong — and because it’s fast and plausible, that wrongness spreads. The habit here is suspicion of the smooth answer: ask whose perspective is baked in, whose isn’t, and who tends to pay when the model gets it wrong. For the deeper mechanics of how this happens, see AI & bias at work: who gets designed in?

2. Keep a human in the loop

Use AI as a tool, not an oracle. Its output is a first draft, never the final word — and the habit to build is the pause between generating and sending. Read it with your own judgement, check it against what you actually know, and own the decision yourself. This matters most when the output touches a real person: a shortlist, an assessment, a reply to someone who’s upset. When AI starts drafting the messages that shape conflict, things escalate fast — keeping humans in the conversation explores exactly that.

3. Be open about using it

Transparency is half of doing this well. You don’t need a disclaimer on every email, but where AI has shaped something that affects people, say so plainly. Hidden, unchecked use is how AI does its quiet harm — and openness does the opposite. It lets people trust the work, and it lets colleagues learn from how you do it. Thoughtful practice only spreads when it’s visible.

4. Design for the edges

The real test of inclusive AI is who it serves at the margins, not the average. Does the output read clearly for someone using a screen reader? Does it assume a name, a culture, a body, a way of working that not everyone shares? Used well, AI is a brilliant accessibility aid — it can open doors that were closed. Used carelessly, it narrows them. Ask “who might this leave out?” as routinely as you ask “is this right?”. More on that in AI, belonging and the future of work.

5. Care for people’s data and dignity

Behind a lot of AI inputs sit real people — their words, their details, their stories. Treat that data the way you’d want your own treated: only share what you genuinely need, think about consent before you feed something personal into a tool, and never let efficiency cost someone their dignity. This isn’t legal advice, and your organisation will have its own rules to follow — but the human instinct underneath the rules is simple. Would the person whose data this is feel respected by what you just did?

6. Model good use

The last habit is the one that multiplies the rest: use AI well yourself, and help the people around you do the same. That means showing your working, being honest about what AI can and can’t do, and bringing colleagues alongside rather than racing ahead. A real divide is opening up between those who can use these tools with judgement and those left behind — and closing it is everyone’s job. The AI readiness divide unpacks why, and using AI for good at work shows what good, inclusive practice looks like in action.

From habits to a way of working

None of this asks you to become an expert overnight — that pressure is exactly what keeps people frozen. Pick one habit, build it until it’s automatic, then add the next. AI amplifies good or harm; you choose which, one small step at a time. When all six pull the same way — you see the risk, check the output, say so plainly, design for the edges, guard the data, and bring people with you — AI becomes a tool that widens the door rather than narrowing it.