The rush to add AI isn’t the same as knowing where it belongs.
I’ve written separately about what AI can and can’t do, but this article is about a different question: what should stay human?
It can help write emails, summarise meetings, search documents, tidy up notes, draft replies, analyse information and suggest next steps. In some cases, it can save a lot of time. In others, it can help someone get started when they are staring at a blank page.
I don’t think businesses should ignore that.
But I also don’t think the right answer is to add AI to everything just because we can.
The question I keep coming back to is not:
How do we add AI to more things?
It is:
Where can AI genuinely help, without taking away the judgement, responsibility and control that good businesses still need?
There is a difference between using AI to make work easier and letting it creep into decisions that still need a person behind them.
It can help with first drafts, rough notes, long documents and finding the right words, especially when the person using it already understands the subject.
But it shouldn’t become a way of avoiding responsibility.
- An AI-generated reply to a customer is still the business’s message.
- A summary of a complaint still needs someone to understand the detail and decide what is fair.
- A suggested answer still needs to be checked by someone who knows whether it is right.
- If AI is involved in pricing, hiring, customer service, advice or anything sensitive, the business still needs to know who checked it, who understood it and who is accountable for the outcome.
In many businesses, trust is not built by the tool being used. It is built by the person who understands the situation, notices the detail, and takes responsibility for what happens next.
I’m not against using better tools. But I would be wary of handing over the parts of a business that customers trust people to get right.
This is where the problems start.
- A member of staff copies private customer details into the wrong tool without thinking.
- We have already covered the danger of putting data at risk when staff use AI tools without clear boundaries, but data is only one part of the issue.
- A confident answer is accepted because it sounds right.
- A message goes out that technically says the right thing, but misses the tone completely.
- A decision is influenced by something nobody can properly explain.
- A business starts to rely on a tool before agreeing where the boundaries are.
None of that means AI should be avoided.
It means businesses need to be more deliberate.
Before using AI in a new area, I think there are some simple questions worth asking.
- What information are we giving it?
- Could that information identify a customer, a member of staff or something commercially sensitive?
- Is the output being checked by someone who understands the subject?
- Would a customer expect a person to be involved here?
- Could this affect someone in a meaningful way?
- If the answer is wrong, who carries the responsibility?
Those questions aren’t technical. They’re business questions.
That’s why the conversation around AI needs to stay practical. Not every business needs a huge policy document before anyone can use a writing assistant. But every business does need some common sense about what is acceptable, what needs checking and what should stay with people.
There are plenty of places where AI can help. But I would be much more cautious where the outcome depends on judgement, trust, context, fairness, tone or responsibility.
The important thing is that businesses stay in control of the tools they use.
So before we add AI to everything, the better question is this:
What should AI help with, and what should stay human?