Before You Let AI Cut a Step From How You Work, Ask This

"Step four adds nothing. I'd cut it," the AI tells you.

You asked AI to tidy up a task, maybe how you send invoices, how you pack for a trip, or how you put the month-end report together. AI came back sounding sure of itself, and step four got the chop.

What if the AI is right?

Maybe it is. Step four has always been a chore. But, honestly, can you remember why step four is there?

Stop for a moment, because "I can't remember why" and "there is no reason" feel the same in your head, but they are different things.

GK Chesterton told a short story about this in his 1929 book The Thing. A man who wants to change things finds a fence built across a road, and because he can't see any use for the fence, he wants it gone. Then the wiser person answers him.

"If you don't see the use of it, I certainly won't let you clear it away. Go away and think.

Then, when you can come back and tell me that you do see the use of it, I may allow you to destroy it."

Why can't the AI see the reason?

Because the reason was never in what you typed. AI is good at steps and will spot repeats and delays, and that's worth asking for.

What AI can't see is the day something went wrong. Maybe a payment went to the wrong account, and that's why you now read the account number back. Maybe the passport was in the other bag on holiday, and that's why there's a second check at the door. You remember those days, or someone else does.

When AI calls a step pointless, it really means nothing you told it explains the step. So treat that as a question, and put the scissors down for now.

How do I check before anything goes?

Pick one task and give it half an hour. First, ask AI for the reason behind each step before anything is cut, which turns AI from a tidier into a detective.

Here is how I do [task], step by step. [your steps]. Don't suggest removing anything yet. For each step, tell me what it might be protecting against. Tell me what could go wrong without it. Then ask me what I know about why each step is there.

For any step you still can't explain, find out who built the fence. That could be a colleague, whoever did the job before you or a family member, or it could be you, three years ago, in an old email.

I can't explain these steps. [list]. Help me work out who might know why each step exists. Then write one short, friendly question I could send them. I'll send it myself.

Then decide, and cut on purpose. Sometimes the reason has gone because the supplier changed, the software now does that job, or the children have grown up. If so, remove the step and write down why.

If you found no reason and not much rides on the step, drop it for a fortnight and see what happens. If a lot rides on the step, keep it until you know.

Sometimes the way you work is too tangled to write out as steps. Then let AI interview you first, until AI understands what you do and why.

What if there's nobody left to ask?

Sometimes the reason is lost for good, and then you're deciding without all the facts, which is normal life. Think about what breaks if you're wrong, and choose.

Don't let AI fill the gap, because AI's guesses about why a step exists can sound convincing and still be made up. Take those guesses to a person, as questions.

Aren't plenty of steps pointless anyway?

Yes, rules often last longer than their reasons, and "we've always done it this way" has wasted more hours than any AI will. If you had to trace the history of every step before cutting it, nothing would ever change.

Chesterton agreed with more of that than people think. Read his rule to the end, where he says "I may allow you to destroy it." All he asked was that you know what you're removing.

The cut is yours, but first do what the wiser person asked.

"Go away and think."