What AI Has Turned Marketing Into (Part 1)

I've been in marketing for nearly twenty years.

In those twenty years, the tools have changed many times. From newspapers to portals, from portals to social media, from social media to short video. Every time it changed, someone said "marketing is about to be revolutionized," and then marketing remained marketing — only the ad slots changed, the creative formats changed, the metric names on the report changed.

This time feels different.

Not because AI is impressive. Because what AI is changing is not the tools — it's the structure.

Conversation, from metaphor to fact

In 2008 I wrote on my blog: "Marketing is a conversation."

Back then, that sentence was a metaphor. The brand spoke, the audience listened, and occasionally someone left a comment or filled out a survey — that counted as "interaction." Conversation was a romantic narrative we wanted to believe in; in practice it was closer to broadcasting — one-way transmission, hoping someone was receiving.

Today, the conversation is literal.

Consumers are chatting with a brand's AI support, interacting with recommendation engines, dealing with the personalized feed they open on their phone every morning. With every interaction, the system is learning, and the next line is being adjusted in real time. Marketing has shifted from "designing a message" to "designing a logic of conversation" — not writing a script, but designing a set of rules that can handle countless possible directions.

A brand is no longer just what it says, but how it responds.

This change sounds like an upgrade, but it's also a transfer — marketers have lost control of the "final version." You can proofread an ad. You can't proofread a conversation.

Narrative, from fixed version to dynamic generation

In the old days of marketing, there was something called the "key visual." One image, one line, used across every placement; everyone saw the same story.

In that era, telling a story meant: you polished it, you signed it off, you shipped it out, and then you prayed it worked.

Today, the story is generated.

Same brand, same product — AI can tell one version to a 25-year-old woman in northern Taiwan and a different version to a 45-year-old man in central or southern Taiwan. Not manually cut variants, but assembled by the system in real time, based on his browsing history, purchase habits, and dwell time, deciding which side of which story to tell him today.

This is the precision marketers have dreamed of.

But one thing nobody says out loud:Before a consumer encounters your story, they may have already asked AI about it.

He asks ChatGPT whether this brand is trustworthy, and AI gives him an answer. That answer is the first layer of narrative he encounters. Your key visual, your brand site, the ads you paid to place — those are the second layer.

What marketers are competing for now isn't just media placements, but AI's database of impressions.

It's a game most people haven't realized they've already lost at the starting line.

Scene, from visible design to imperceptible environment

"Marketing guides people into a pre-designed scene."

I've been writing that line for almost twenty years. The "scene" used to be visible — store flow, ad landing pages, exhibition design, unboxing experiences. You could sense someone was designing you, and even when you were being designed, you knew where you were.

Today, the scene is an environment built by algorithms.

An informational reality constructed by personalized push notifications. Everyone lives in a bubble tailor-made for them, yet feels no guidance — because everything seems to "just match my taste." It's not coincidence. It's design. Only the designer is the system, not some adman you can name.

The designed scene has become a place you can't walk out of, and don't know you're inside.

So what has AI changed?

It has made conversation real, narrative dynamic, and the scene imperceptible.

The marketer's range of capability has expanded. Reach is more precise, messages more personal, scenes harder to escape.

This is what marketers have wanted for twenty years.

In Part 2, I want to talk about this: is there a problem that, from 2008 until now, no evolution of AI has ever solved?

Further reading: - One Problem AI Can't Solve (Part 2) - How Satsuma Makes Advertising - Big Idea vs Playing It Safe