I stopped using AI… kind of.

I started to feel like my thinking and problem-solving were getting diluted because I was using AI for everything.

A problem came up, I opened AI. Someone asked me a question, I opened AI. I needed to work through an idea, and I was halfway through a prompt before I’d given myself a chance to think.

That last one bothered me.

I was getting answers, but I was increasingly skipping my own first attempt. The messy bit where you don’t quite understand the problem yet, question what you’ve been told, and try to work out where to start.

So I’ve started an experiment. I still use AI. I’ve changed when I invite it in.

Going old skool 🤣

When a problem or enquiry comes my way, I’m trying to ask myself first.

What do I think is happening? Why is it happening? Do I understand the problem well enough to start looking for an answer?

Then I go old skool and use the five whys.

It gives me a reason to pause rather than immediately reaching for a solution. Asking “why?” again often exposes something I’ve accepted without checking.

Take a hypothetical request: “We need to automate these customer updates.”

It would be easy to open AI and ask for an automation workflow. But there are a few questions worth answering first:

• Why do the updates take so much time? Because someone has to chase the information.
• Why do they have to chase it? Because it’s spread across different places.
• Why is it spread across different places? Because each team records progress differently.
• Why does each team do it differently? Because there isn’t an agreed handover.
• Why isn’t there an agreed handover? Because nobody clearly owns that part of the process.

Those answers would need checking with the people involved. Five whys can turn into five guesses if you’re not careful, and some problems have several causes.

But even that initial questioning changes what I’d investigate. I might still need automation. I’d also need to understand the ownership and handover problem before building anything.

Having a go before asking for help

Once I’ve questioned the problem, I try to think it through myself.

What would I do with what I currently know? What options can I see? What am I assuming? Where does my explanation stop making sense?

I want to get far enough to have a view of my own, even if it’s incomplete or wrong.

There’s a difference between recognising an answer that sounds sensible and being able to explain how you arrived at it. I want to keep practising the second one.

That doesn’t mean sitting there indefinitely to prove I can manage without help. Getting stuck is useful if I can identify what I’m stuck on.

“I don’t know how to solve this” is a starting point. “I can see two approaches, but I don’t know which fits this constraint” gives me something much more specific to investigate.

Then comes the research

When I reach that point, I go back to the usual research questions:

Who does this already? Who do I think might have solved something similar? Is there a better way?

That might mean reading documentation, looking at an existing product, finding a case study, or speaking to someone who deals with the problem.

I also want to understand why their approach works. A solution from another business can look convincing until you realise they have different customers, resources or constraints.

This is still part of forming my own view. I’m looking for evidence, including anything that tells me my first idea was wrong.

Only then do I bring in AI

By this stage, I have something to bring to the conversation.

Instead of asking, “How do I solve this?”, I can say:

Here’s the problem as I understand it. These are the assumptions I’ve made, the approaches I’ve considered and what I found in my research. This is where I’m stuck. What am I missing, and what evidence would help me test my thinking?

That’s the role I want AI to play: a partner I can work through the problem with.

It can challenge an assumption, suggest another approach or help me investigate a gap. I still need to check what it gives me, and I need to be willing to change my mind when there’s a good reason.

An experiment, not a victory lap

I’m not claiming this has suddenly made me a better thinker. I’ve started an experiment because I noticed a habit I wasn’t comfortable with.

And yes, I appreciate the irony of using AI to help shape a post about this. Writing down my experience and using help to communicate it fits the boundary I’m trying to set.

What I want to find out is whether I can explain my decisions more clearly, spot my assumptions earlier and be more specific about the help I need.

For now, whenever a problem or enquiry comes in, I’m giving myself the first attempt. Then I research where I’m stuck. AI joins after that.