AI Isn’t the Problem. Where You Put It Is.

The smartest AI strategy may be knowing when not to leave it alone.

There is a strange thing happening in business right now.

Companies have been handed one of the most remarkable new employees they have ever had.

It can research. It can organize. It can analyze. It can brainstorm. It can write. It can help solve problems at astonishing speed.

And some companies seem determined to put it behind the complaint desk.

Alone.

That last word matters.

Because AI absolutely can work in customer service. It can work in a creative department. It can work in accounting, marketing, research, operations and almost anywhere else we choose to put it.

The problem isn't necessarily where we're putting AI.

It's when we decide AI no longer needs a human sitting beside it.

I use AI. A lot.

I think it is extraordinary.

It has become part of the way I work—not because I want it to think for me, but because it can help me think better.

I can bounce an idea against it, challenge an assumption, ask it to argue the other side, research something I don't know, or take an idea somewhere I hadn't considered.

Used that way, AI doesn't diminish human thinking.

It can expand it.

But lately I've been watching companies use the very same technology in places where its efficiency may be exactly the problem.

THE SAME EMPLOYEE. TWO VERY DIFFERENT JOBS.

Imagine hiring one extraordinarily capable employee.

Put that employee beside a designer, engineer, researcher, accountant, writer or strategist and say:

Help this person do better work.

That employee can be transformative.

Now take exactly the same employee, remove the human sitting beside it, and put it between an angry customer and a company that has made a mistake.

Tell it:

Make this go away as cheaply and efficiently as possible.

Same employee.

Very different outcome.

And that's where I think much of the conversation about AI has gone wrong.

We keep asking whether AI is good or bad.

Maybe that's the wrong question.

Maybe we should be asking not only where we put it, but what—or whom—we remove when we do.

EFFICIENCY ISN’T ALWAYS THE JOB.

When a customer contacts a company because something has gone wrong, the objective shouldn't simply be to close the ticket.

The objective is to repair the relationship.

Those are not the same thing.

A customer may have lost money. They may be frustrated. They may feel ignored. They may have spent hours trying to fix something the company caused in the first place.

Then comes the perfectly polished response:

We understand your frustration.

We value your feedback.

Thank you for your patience.

Your experience is important to us.

Every sentence is technically appropriate.

And somehow the entire thing can feel like nobody was there.

AI didn't necessarily fail.

It may have performed exactly the job it was given.

The human decision about what that job should be may have failed.

BUT WHAT IF AI SAT NEXT TO THE HUMAN?

Imagine that same unhappy customer reaches a real customer-service representative.

Before that human has finished saying hello, AI could already have found the customer's previous emails, identified how many times they've contacted the company, summarized what happened, pulled up what was promised and surfaced the information needed to solve the problem.

AI just did in seconds what might have taken the employee ten minutes and three computer screens.

Now let the human do the part humans are particularly good at.

Listen.

Understand the frustration.

Recognize the nuance.

Make a judgment.

And decide what would actually make things right.

AI supplies the information.The human supplies the judgment.Together, they provide better customer service.

That's not replacing customer service with AI.

That's using AI to make a human better at customer service.

And the same principle works in a creative department.

AI can research. It can challenge an assumption. It can generate possibilities, find connections and give an idea something to bounce against.

But the creative still decides what is good.

That's the partnership.

AI doesn't have to replace the employee to transform the job.

Sometimes its most powerful position isn't in the employee's chair.

It's in the chair next to them.

THERE IS A DIFFERENCE BETWEEN COST AND VALUE.

I understand why companies are doing this.

AI can handle enormous volumes of communication. It doesn't take lunch. It doesn't need vacation. It can respond instantly. It can dramatically reduce the cost of customer service.

Put that number on a spreadsheet and the decision can look brilliant.

But spreadsheets are very good at measuring what something costs.

They're not always as good at measuring what something costs you later.

What is the value of a customer who stays for twenty years?

What does it cost when that customer stops recommending you?

What does it cost when a beautifully automated response makes someone feel that nobody at the company actually cared enough to read what happened?

Those losses don't always appear neatly in this quarter's numbers.

But they're real.

THIS ISN’T AN ARGUMENT AGAINST AI.

Quite the opposite.

I think we're wasting some of AI's greatest potential by treating it primarily as a substitute for payroll.

The exciting question isn't:

How many people can this replace?

It's:

What could our people accomplish if they had this?

Let AI find patterns humans would miss.

Let it research.

Let it organize thousands of pieces of information.

Let it perform the repetitive work that consumes hours without requiring judgment.

Let it give a creative person another direction to consider.

Let it give a customer-service representative the entire history of a problem before they have to respond.

Let it help a human solve the problem.

But when judgment, empathy, taste, responsibility or trust matters, perhaps the human shouldn't disappear from the process.

AI can help write an apology.

A human should know whether the company actually owes one.

AI can identify a customer's problem.

A human should decide what making it right looks like.

AI can generate twenty possible answers.

A human should know which one deserves to be sent.

THE HUMAN STILL ASSIGNS THE DESK.

That's the part I don't want us to lose in all the excitement.

AI isn't walking into companies, firing employees and choosing its own office.

People are deciding where to put it.

People are deciding what authority to give it.

People are deciding when a human gets involved—and when one doesn't.

And perhaps that's where we've been thinking about the office all wrong.

AI can have a desk in every department.

It can work in customer service.

It can work in creative.

It can work in finance, research, marketing, operations and places we probably haven't even imagined yet.

But giving AI a desk doesn't mean we have to take somebody else's desk away.

AI can be on the team. It shouldn't be left alone in the office.

So when AI produces a wonderful result, there is usually human judgment somewhere in that story.

And when AI produces something tone-deaf, shortsighted or absurd, there is often human judgment somewhere in that story too.

Maybe the future isn't about choosing between humans and AI.

Maybe it's about becoming much better at deciding how they work together.

Because AI may be one of the most capable new employees we've ever hired.

But somebody still has to know the customer.

Somebody still has to recognize the good idea.

Somebody still has to care about what happens next.

And somebody still has to be the creative director.

That job is ours.