We bought artificial intelligence to save everybody time.
Excellent. Where is it?
Because I keep hearing about the hours we’re saving, and I’d like to meet somebody who got theirs back.
Did they go home early? Eat lunch sitting down? Get through a Friday without inheriting a problem that apparently became urgent at 4:47?
Or did we discover they had room for another job?
Apparently this is normal.
The numbers sound familiar
Korn Ferry released a workforce report on September 8 based on a survey of more than 16,000 professionals across 11 markets. Sixty-one percent said they were carrying responsibilities from more than one role. Sixty-two percent reported significantly heavier workloads over the past two years. Source: Korn Ferry
Those numbers don’t prove AI caused the extra work. But they ought to make us ask a few questions before the next presentation about all the time technology is giving us back.
I’ve spent enough years around railroad operations to recognize the distance between a plan and the people expected to make it work.
On a slide, the process has six steps.
On the floor, step three requires finding somebody who knows why step two doesn’t work on nights.
That person is usually busy.
We should probably put that in the presentation.
I use AI. I still have questions.
I use these tools myself. They can help me organize ideas, work through a problem, and get something done that would otherwise take longer.
There’s value in that.
But saving time on a task and reducing somebody’s workload are two different accomplishments. You can do the first without ever doing the second.
Say a tool cuts an hour of work down to twenty minutes.
Now we have forty minutes to account for.
We could use some of it to check the result. We could catch up on work that’s been waiting. We could give somebody enough room to think before the next decision.
Or we could immediately assign another hour of work.
Then put “efficiency gains” in the monthly update.
The tool did its part. Somebody else decided what the worker got out of it.
The work after the work
Here’s the part I want counted.
Who checks what the system produces?
Who catches the wrong information? Who fixes the exception? Who explains the result to a customer, a crew, or a manager who assumes the computer already handled it?
Somebody still has to know enough to recognize when the answer is wrong.
That knowledge took time to build. It came from doing the work, making mistakes, asking questions, and remembering what happened the last time somebody said, “That should be fine.”
Now we’re asking that person to do their regular job and supervise the output of a system we purchased to reduce their workload.
Maybe it still saves time. Good. Measure the whole job and find out.
Include the checking. Include the corrections. Include the training and the days when the thing everybody depends on decides it needs an update.
If we’re counting the savings down to the minute, we can count the cleanup, too.
The reliable person will make it look successful
This is where I get concerned.
The dependable person will usually find a way.
They’ll learn the new tool. They’ll work around its limitations. They’ll catch the mistake before it travels far enough to become embarrassing.
And if nobody asks how they managed it, the result can look like proof that the new arrangement works.
The report got finished.
The deadline got met.
The customer got an answer.
What doesn’t show up is the lunch they worked through or the other task they pushed into tomorrow. Eventually tomorrow fills up, too.
You can run that arrangement for a while. Reliable people can keep a strained system looking healthy longer than you’d expect.
That doesn’t make it healthy.
It makes the strain harder to see.
Managers have to account for what they add
I’ve been the person responsible for getting the work done. I understand the pressure to find capacity where there isn’t much left.
That’s exactly why I think leaders need to answer a plain question whenever they introduce a new tool:
What are we taking off the person’s plate?
Name it.
If the answer is a task that takes an hour, find out whether it actually takes an hour after implementation. Ask the person doing it. Watch the process. Come back after the launch excitement wears off.
And if the plan is to use every saved minute for additional work, say that clearly.
People deserve to understand what’s being asked of them. They can handle a straight answer better than a cheerful announcement about making their lives easier while their assignment list gets longer.
I’d also ask who owns the mistakes.
Giving somebody responsibility for checking a system means giving them time to check it—and enough authority to stop the process when something looks wrong.
Otherwise, we’ve handed them accountability with a login screen.
I’d like to see the time show up somewhere
A useful tool should make a meaningful difference.
Maybe that means fewer errors. Maybe it means a crew gets the information it needs sooner. Maybe a supervisor finally has time to teach somebody instead of answering questions while walking toward the next problem.
Those improvements matter.
So does finishing a shift with enough left in you to be decent to the people waiting at home.
I’m interested in technology that helps us get there. I’d like to see organizations measure those results with some of the enthusiasm they bring to counting output.
Because when somebody tells me we’ve saved a thousand hours, my next question is going to be where those hours went.
If the answer is that we eliminated a position and distributed the remaining work among everybody else, then explain that decision.
But don’t tell the person carrying two jobs that you’ve given them their time back.
They know what time they got home.
Has a new tool actually made your job easier—or just changed how much you’re expected to get done? I’d like to hear what happened where you work.