Earlier posts in this series looked at what generative AI does and does not do well, and at some of the ethical and practical questions it raises. This post turns to a more immediate question: if your organisation or team wants to engage with AI more deliberately, where do you actually start?
There is now a growing range of guidance on AI ethics, governance, and responsible use. This can help clarify what needs attention. But it often leaves a practical gap: what are the first conversations to have, and who should be part of them?
What follows is a simple starting point. It outlines five areas that organisations tend to work through, and some questions that can help get those conversations going.
Seeing the overall shape
When organisations begin working with AI, a few areas usually come into view:
- who is paying attention and providing direction
- how teams are already making sense of AI in their work
- where AI could be useful, and where it could cause problems
- what shared expectations about use and limits are needed
- how the organisation keeps learning as things change
These areas apply whether you are thinking about AI across an organisation, within a programme, or as a team. The scale may differ, but the conversations are similar. These do not need to be addressed all at once, or in a fixed order. But they help show the overall shape of the work. If you are just getting started, it is often enough to pick one or two of these areas and begin with a small conversation. You do not need a full plan in place.
1. Getting the right people engaged
This is often where things start. Leadership takes an active interest in how AI is already being used, and brings together a small group to look at it from different angles. In many organisations, AI tools are already in use informally. Starting with what is happening, rather than what should happen, tends to ground the conversation. In many cases, this can be as simple as a short discussion with a few people who are already using these tools.
A question to start with: Where are we already using AI, even informally, and who needs to be involved in thinking about it?
Things worth exploring:
- What are people using AI for, drafting, analysis, summaries, something else?
- What has felt useful, and where has it felt uncertain?
- Who understands the work context well enough to see where AI matters? Who understands the risks or obligations? Who is already experimenting?
Getting these people into the same conversation early makes a difference. It does not need to be a formal committee, a small working group with a clear brief is usually enough.
2. Making sense of AI with teams
Once there is some leadership attention, the next step is often to create space for teams to think about what AI means for their own work. This matters because AI tends to show up in specific tasks, decisions, and relationships. The people doing the work are usually best placed to see where it could help, where it could cause confusion, and where they would want to keep human judgement in place.
A question to start with: How might AI change how we do our work, for better or worse?
Things worth exploring:
- Where are the time pressures or bottlenecks where AI might be genuinely useful?
- Where do we rely on interpretation, relationship, or contextual judgement?
- What might change, in the work itself, or in how it is received, if AI were introduced?
These conversations often surface concerns that would not come through in a top-down process. They also tend to build a more realistic picture of where AI actually fits.
3. Clarifying boundaries and developing shared guidance
As the picture becomes clearer, most organisations find they need some shared expectations about how AI is used. This does not have to be a formal policy from day one. But it does mean working through a few key questions together.
The harder conversations tend to sit here. Questions about what is acceptable to share with an AI tool, what needs to be checked before use, where the organisation would draw a line, and what kinds of risks need to be managed are not always straightforward. Different parts of the organisation may see these differently, and that is worth surfacing rather than trying to resolve everything at once.
A question to start with: What would good use and poor use of AI look like in our context?
Things worth exploring:
- What would we be comfortable with if it were visible externally?
- What should be reviewed or checked before use?
- Who is responsible for AI-assisted outputs?
- When should AI use be disclosed?
- What kinds of risks are we most concerned about in our context?
- What do people need to know to use AI responsibly here?
The aim is to capture these decisions in a form people can actually refer to, something light and usable, not a document that sits unread. It will need revisiting as experience grows.
4. Regular review and learning over time
AI tools and their uses will keep changing, and so will the organisation’s understanding of where they help and where they do not. Building in some form of regular review, even lightweight, helps the organisation stay responsive rather than locked into early assumptions.
This is also where the quality of organisational learning shows. It is relatively easy to adopt AI tools. It is harder to notice when something is not working well, when a boundary needs adjusting, or when early enthusiasm has outrun careful use.
A question to start with: What are we noticing from using AI, and what might we change?
Things worth exploring:
- What has worked better than expected?
- What has not sat right, and why?
- What would we want to adjust, in how we use AI, or in the guidance around it?
- Are there new uses emerging that we had not anticipated?
Over time, this kind of review helps build a more considered approach to AI use. It makes it easier to see where these tools are genuinely helpful, where they introduce risk, and how practice needs to adjust as experience grows.
Closing reflection
Getting started with AI in an organisation is less about putting a complete system in place at the outset, and more about building shared understanding through use.
In many settings, the quality of judgement around AI, how it is used, interpreted, and questioned, matters as much as the technology itself. These starting points are not a complete model. They are a way to begin, in a way that keeps learning and context in view as experience builds.
This post is part of a small set of reflections on using AI in practice, from individual use through to shared work and wider ethical considerations. You can explore the full set on the AI for reflective practice, research, and collaboration hub. A few related posts from that collection include:
- AI as a thought partner: reflections on collaborative practice and systems work
- AI prompts for shared thinking: a light framework for purposeful prompting
- Working with AI in the room: authorship, responsibility, and collective judgement
- Seeing the wider ethical picture around AI development and use
- AI in place-based practice: what is shifting
The hub also links through to related resource pages, including Using AI in research and practice and and AI in context: the wider picture.
[* Image by Jintana / Adobe Stock]
