AI Should Stop Racing on Efficiency — and Start Racing on Understanding People

A few days ago I came across a ranking that lined up the companies doing the best work with AI in marketing this year.

I had assumed it would be the usual story. AI writing copy, AI generating images, AI cutting ad spend.

But after I read through several of the cases, I paused.

Almost none of these companies were using AI as a cost-saving tool. They were using it for one thing: to figure out what users actually wanted — one step earlier than the users themselves.

That's interesting.

What does "one step earlier than the user" mean?

Let me start with a phone maker.

By the usual logic, marketing a phone only begins the moment a user starts searching for one. They walk into a store, compare prices, read reviews. That's the agreed starting point of the sales funnel.

But this company noticed something: before users ever search for a phone, their behavior is already leaving traces. Maybe an app is getting noticeably slower. Maybe they've suddenly started watching comparison videos. Maybe their shopping cart has picked up a few items that have nothing to do with phones.

Looked at individually, none of these actions have anything to do with buying a phone.

But when AI stitched them together, the signal was clear: a latent demand was taking shape.

So marketing reached them earlier — and converted better than blasting ads ever could.

This is advertising shifting from "waiting for users to come to us" to "anticipating which door the user will walk through next."

Now let me tell you about a car company

Even more interesting is a luxury car brand.

Their logic is simple: no one wakes up in the morning and suddenly searches "I want to buy a car."

Buying a car is something life pushes you into. Moving house, changing jobs, having a baby, getting promoted.

Every one of these life events leaves traces all over e-commerce platforms. A crib purchased, school districts in a new city searched, new lifestyle content subscribed to. Long before the car has even entered the user's mind, these behaviors are already pointing to one conclusion: "This person may be in the market for a new car."

What AI does is read these seemingly unrelated behaviors and detect that "relevance has emerged."

Marketing is no longer about interrupting users — it's about showing up at exactly the moment they need you.

A bank that can recognize "diligence"?

What surprised me most was a bank.

After a merger, the usual playbook is to segment customers by demographic tags. Age, income, occupation — then blast them with product information.

But this bank took a different angle.

They used AI to look through anonymized behavioral data, searching for one thing: diligence — a virtue the culture prizes.

What does diligence mean here?

Paying credit cards on time, long-standing stable savings habits, sustained care for a family account. These small consistencies hidden in transaction records — the bank identified them and translated them as: "This person is taking their life seriously."

And then marketing stopped being about pushing products. It became a way of saying, to people who take life seriously, "I see you."

When I read this case, my heart skipped a beat. A financial institution using AI — not to sell more precisely, but to recognize a cultural value. This is what AI should be doing.

The common thread across these cases

Let me quickly walk through a few more.

An oil brand used AI in Formula 1 to generate, for every single fan, a race-track story that belonged only to them — turning a can of fuel into an emotion.

An airline, in the seconds when a user was hesitating over a booking, used AI to assemble a creative idea that matched the precise spirit of this trip. Not a discount — value.

An e-commerce platform used AI to read the business state of small business owners in Taiwan, judging who was ready to go global — and then told them how to take the first step.

A maternal and baby brand built a model called "Mommy Logic" that delivers information and support at the moments when mothers are most anxious and most uncertain — instead of pushing products.

There's even a health-supplement brand that created an AI companion role, just to talk to young people who are lonely in big cities.

Do you see it now?

In every one of these cases, AI isn't helping brands sell more. It's helping brands better understand what the person on the other side is thinking right now.

After finishing this ranking, a few observations

First observation: the marketing industry is quietly changing its subject.

For the past decade the subject has been "traffic," "conversion," "ROI."

Now the subject is slowly becoming "signals," "needs," "relationships."

AI isn't there to help you shout louder. It's there to help you listen more clearly.

Second observation: the AI applications that really work have a remarkably "soft" quality to them.

Not cold algorithmic black boxes — they take genuinely fuzzy things like cultural values, emotional moments, and major life changes, and turn them into something you can respond to.

Diligence can be recognized. Hesitation can be seen. Major life events can be anticipated. These used to be things marketers could only guess at by rule of thumb — now AI helps you catch them more accurately.

Third observation — and this is the one I most want to say:

Many companies are still racing each other on "how much money can AI save me." But a group of companies has already started racing on "who can AI help me understand better."

These two paths are diverging, more visibly every day.

Keep walking the first path, and you can only compete with rivals on who's cheaper.

Walk the second path, and you might just compete with rivals on who matters more to people.

I don't know the answer. But the direction is worth thinking about.