The advertising industry finally found the edge of the AI hype pool and, thank God, some people are climbing out.

For the last two years, every marketing conference sounded like a robot karaoke bar. AI will write the ads. AI will make the videos. AI will pick the audience. AI will personalize the funnel. AI will replace the agency. AI will optimize your socks off while your team sips espresso and says “brand ecosystem” with a straight face.

Cool story.

The hotter take coming out of Cannes this year is much more useful:

AI will not save advertising. It will expose bad advertising faster.

That is not anti-AI. It is anti-delusion. There is a difference.

Cannes got a little less drunk on the future

The latest round of Cannes marketing chatter has been weirdly refreshing. In a Decoder conversation with Digitas North America CEO Amy Lanzi, the point was not “AI will magically replace creativity.” It was closer to: AI helps operations, systems, and scale, but the ad industry still needs actual ideas.

Business Insider’s Cannes coverage landed in the same neighborhood. At an AI marketer roundtable, leaders were not just yelling about speed. They were talking about taste as a competitive advantage, because when everyone can produce polished content, the rare thing is not output. The rare thing is judgment.

That should scare the hell out of mediocre marketing teams.

Because if your advantage was “we can make a lot of pretty assets,” your advantage is gone. If your agency pitch was “we move fast,” congrats, the machine moves faster. If your brand has no point of view, no product truth, no distribution discipline, and no taste, AI is not going to give you a soul. It is going to give you 400 variations of the same dead-eyed campaign.

AI makes the average stuff look better, which makes it more dangerous

Bad advertising used to look bad.

That was helpful. You could spot the ugly landing page, the clunky email, the stock-photo ad, the painfully generic copy. The stink was visible.

AI changes that.

Now bad advertising can look competent. It can have clean grammar, decent layouts, slick video edits, synthetic product shots, fake polish, and a headline that sounds like it survived three rounds of committee feedback.

That does not make it good.

It makes it harder to diagnose.

The new failure mode is not “this looks amateur.” The new failure mode is “this looks professional and says absolutely nothing.” That is worse, because teams mistake polish for progress. They point at the dashboard and say the content machine is working because the machine is busy.

Busy is not strategy.

The output race is a trap

Gartner recently said marketing leaders expect AI-driven automation of marketing work to jump from 16% in 2026 to 36% by 2028. That number matters, but not because it means every marketer should panic-buy another tool.

It means the baseline is rising.

More briefs will be drafted by AI. More campaigns will be assembled by AI. More reports will be summarized by AI. More creative variants will be generated, tested, killed, and generated again before the second coffee.

That is fine. Some of it is great. A lot of marketing work is repetitive as hell and deserves automation.

But once everyone has the same speed, speed stops being the advantage.

The advantage becomes knowing what should exist in the first place.

What is the idea? What is the tension? What does the buyer already believe that you need to break? What proof do you have? What language belongs to your brand and what language makes you sound like a LinkedIn ghostwriter trapped in a SaaS booth?

AI can help you execute the answer.

It cannot care enough to find the answer for you.

Creators are eating the lazy middle

There is another reason the “AI will save ads” fantasy feels so weak right now: creators are showing brands what attention actually looks like.

Creators are not winning because they have better render settings. They are winning because they have taste, trust, timing, weirdness, specificity, and a relationship with the audience. They know what feels dead on arrival. They know when a line sounds fake. They know when a trend is already cooked.

Most brand teams want that energy but keep sanding it down until it becomes campaign foam.

Then they ask AI to scale the foam.

That is how you get 19 assets, 42 captions, 6 email variants, and zero pulse.

The uncomfortable lesson from creator culture is that people reward recognizable humans, not optimized rectangles. Even when AI enters the workflow, the advantage is still direction. The person who knows what to keep, what to cut, what to risk, and what to leave rough.

Taste is not decoration. Taste is decision-making under pressure.

The real AI marketing stack is not a content cannon

The dumb version of AI marketing is obvious:

“Make more ads.”

The useful version is quieter:

1. Watch the market.
2. Pull customer objections out of support, reviews, sales calls, and search behavior.
3. Turn real pain into sharper campaign angles.
4. Generate options.
5. Have humans with taste choose and refine.
6. Push winners into assets, product pages, dealer content, sales decks, and paid campaigns.
7. Measure what actually changed.

That is not a content cannon. That is a feedback loop.

And feedback loops need infrastructure.

If your product images are scattered, your specs are stale, your dealer data is a mess, your MAP policy is getting ignored, and your proof lives in someone’s desktop folder named “final_assets_use_this_one,” you do not have an AI marketing problem.

You have an operational hygiene problem.

AI will happily amplify that mess.

Brand systems beat brand vibes

This is where a lot of teams need to grow up.

Brand is not just voice. Brand is not just a logo. Brand is not just a campaign platform cooked up in a deck with moody photography and the word “human” used 47 times.

Brand is the operating system for how your company shows up.

It includes the claims you make, the proof you can defend, the assets you approve, the partners you trust, the retailers you support, the prices you enforce, the product information you publish, and the customer questions you answer before the buyer bounces.

That is why tools like ToughAssets, ToughMAP, and ToughLocator matter in the AI era. Not because software magically creates taste. It does not. But clean assets, pricing discipline, and location truth give your marketing machine something real to work with.

AI needs context. Buyers need trust. Brands need control.

The companies that treat those as separate problems are going to keep publishing prettier garbage.

What to do before your next AI campaign

Before you ask AI for another campaign concept, do the boring work that actually makes campaigns better.

Audit your last ten ads. Strip the logo off. Would anyone know they came from you?

Read your product pages out loud. Do they sound like a person with conviction or a brochure possessed by SEO software?

Pull your last 100 customer questions. Are your ads answering real objections or just making claims your competitors could copy by lunch?

Open your asset library. Can a marketer, dealer, agency, or AI workflow find the correct product image, specs, claims, and usage rules in under a minute?

Check your pricing and channel presence. Are you building trust, or are you letting the market turn your brand into a scavenger hunt?

That is the work.

Then automate.

Not before.

The take

AI is not the death of advertising. It is the death of pretending.

It kills the excuse that production is too slow. It kills the excuse that personalization is too hard. It kills the excuse that your team cannot test more angles, listen to more signals, or build better loops.

But it also kills the illusion that output equals taste.

The next great marketing teams will absolutely use AI. Aggressively. Daily. Deep in the workflow. They will use it to research, draft, remix, measure, route, and scale.

But they will not ask it to replace judgment.

They will build stronger brand systems, cleaner data, sharper creative direction, and tighter feedback loops. They will use AI to move faster after they know where the hell they are going.

Everyone else is about to discover that bad advertising at machine speed is still bad advertising.

It just gets annoying faster.