There isn’t one ethical answer: what a room of council professionals taught me about AI

At a recent talk about AI ethics, I asked a room of council professionals how they felt about AI. There was curiosity, excitement, exhaustion and fear. That range of reactions told me something useful: we cannot make good decisions about AI by pretending everyone sees the same benefits or carries the same risks.

The conversation that followed moved beyond what AI can do. We asked when it helps a council serve people, when it shifts a problem somewhere else, and who remains responsible for the result. This lesson gives you a way to have that conversation in your own council. The accompanying slides set out the six questions we used to guide it.

Useful does not always mean wise

I use AI in my own work. It helps me organise information, get ideas out of my head and make them easier for other people to follow. As a dyslexic person, that can be transformative. I also know how tempting it is to reach for it when a task feels overwhelming.

But “Can AI do this?” is only the first question. Imagine an officer receiving a long, emotional message from a resident. AI could summarise it in seconds and suggest a reply. That may give the officer breathing space. The resident, though, needs to know someone has understood the concern and taken responsibility for answering it. If the suggested reply sounds considerate but nobody has properly read the message, the council has gained speed and lost trust.

There is useful work for AI in sorting notes, making a first draft or helping someone find the points they need to address. A person still needs to check the facts, make the judgement and decide how to respond.

The room held different values

During the discussion, people raised the environmental and human costs behind AI. Others described the practical difference it can make when someone struggles with a task or a way of working that other people take for granted. Both concerns deserve space. A decision that looks responsible from one angle can make work harder or less accessible from another.

That does not mean a council has to use AI for everything, or avoid it altogether. It means naming the values involved before making a rule. What benefit are we hoping for? What resources does it use? Who gets access to that benefit, and who might carry the cost? Councils can reach different decisions and still take those questions seriously.

Keep the person at the other end in view

We talked about where automated help makes sense. A simple way to report an overflowing bin could save a resident a phone call and help staff send the report to the right place. That is a clear task with a clear next step.

Now think about someone who is confused, distressed or unable to get what they need from an online service. Sending them round a chatbot loop may reduce one type of work for the council while creating another problem for the resident. The question is not simply whether the chatbot works. It is whether the person can still reach a human when they need one.

The same principle applies to staff. If AI removes routine work, what happens to the time it saves? Does it create more room to listen and solve problems? Or does it leave fewer people handling more difficult conversations? That is a choice about how work is organised, as much as a choice about technology.

A decision you can explain

In the slides, I offer six questions to slow a decision down just enough to make it better. You do not need a technical background to use them.

Purpose: What problem are we trying to solve, and is AI the right way to solve it?

People: Who benefits, and whose voice, work or access could be affected?

Data: What information would we need to share, and do we need to share it at all?

Fairness: Could the output repeat an assumption, miss someone or treat people differently?

Planet: Is the value of this use worth the resources behind it?

Accountability: Who will check the result, change it if necessary and answer for the final decision?

Those questions might lead to “yes”, “yes, with safeguards”, or “no, use a simpler approach”. An ethical decision is one you can explain to the person affected by it, including what a human checked and who they can speak to if something is wrong.

Try it with your council

Imagine your council receives a lengthy complaint. An officer wants to use an AI tool to identify the points requiring a response and draft an acknowledgement. Take a moment before you decide whether that is appropriate.

First: What would genuinely help the officer without treating the resident as a block of text to process?

Then: What information would the tool need, what could be removed, and is it approved for that use?

Finally: Who would read the original message, check the summary and take responsibility for the reply?

You may decide that a tool could help organise the work, provided a person reads the complaint, protects the information and writes the response with care. You may decide that this particular message needs a direct conversation. The point is to make the choice deliberately, rather than letting convenience make it for you.

What stayed with me after the talk was not a single answer from the room. It was the value of hearing people bring different experiences to the same question. AI is moving quickly. Our best chance of using it well is to keep talking about the people affected, keep checking what it produces, and keep hold of the decisions that belong to us.