AI wins at clicks. Humans win at meaning. Brands need both.

Human creativity is not in danger from AI. But the AI versus human framing keeps marketing leaders from asking the more useful question. AI-generated content is real, it works in specific contexts and dismissing it on principle is as shortsighted as embracing it wholesale. It is important to understand what AI actually does well, what it does not, and what that means for how brands should be deploying it.

A recent study published by researchers at NYU and Emory University delivered a finding that made a lot of creative directors uncomfortable: fully AI-generated ads outperform human-made ads on click-through rates by up to 19 per cent. The reaction in many marketing circles was alarm; in others, triumphalism. Both responses missed the more useful question.

AI-generated ads, when built from scratch rather than used to modify existing human work, produced stronger visual processing fluency and higher emotional engagement in the moment of exposure. That tells us something important about how AI handles pattern recognition, visual coherence and immediate response optimisation.

The study measured short-term click-through rates. It did not measure brand preference, purchase intent over time, customer loyalty or any of the indicators that determine whether a brand is building lasting value. CTR measures the probability that someone acts on a stimulus in a single moment. That is a narrow proxy for brand strength, but a useful data point for understanding how to make direct-response advertising more effective.

What performance metrics actually measure

Click-through rate rewards predictability. It favours familiarity, visual clarity and pattern-matching against what has already worked. AI excels at this because it is trained on vast libraries of past performance data. Give it a goal, a feedback signal and enough data, and it will optimise toward that goal with a consistency no human team can match. That is exactly why performance metrics should not be the ceiling of creative ambition.

Optimisation follows patterns. The brands that define culture define patterns instead of following them. Hermès has never optimised its way to relevance. The brand’s endurance is a function of refusal as much as execution: the refusal to accelerate production beyond what craftsmanship allows, to expand distribution beyond what exclusivity requires and to respond to trends that would dilute what the brand means to the people who have chosen it for decades. None of those decisions would survive a CTR test. All of them built one of the most valuable brands on earth.

This does not mean brands should ignore performance data. It means they should not mistake a rise in CTR for evidence that the brand is getting stronger.

The long game has different rules

Timeless brands are not built on impressions. They are built on memory, coherence and emotional truth, or the accumulated weight of consistent experience over time. When someone chooses a brand they have not seen advertised recently, pays a premium without comparison shopping, or recommends it without being asked, that is brand equity at work. None of it shows up in a click-through report.

LVMH is instructive here. The group has adopted AI across logistics, inventory and personalisation. But the creative and strategic decisions that define what Louis Vuitton or Dior means to its customers are not delegated to optimisation systems. Those decisions are made by people who understand that the brand’s value is inseparable from its judgment about what it will and will not do. Technology scales the experience; it does not set the standard.

The metrics that track long-term brand strength – like preference, Net Promoter Score, repeat purchase rate and customer lifetime value – are slower to move and harder to attribute to a single campaign. They also reflect something no display ad can manufacture: the quality of the relationship between a brand and the people who choose it repeatedly. Building that relationship requires human judgment applied consistently over time, including judgment about which cultural moments to engage and which to let pass.

Where AI belongs in the system

None of this argues against AI in brand and creative work. The challenge is knowing where AI creates value and where it creates the illusion of progress.

The NYU/Emory study offers a useful insight. AI-generated ads outperformed human ones when given full creative latitude. AI-modified ads, where humans set the direction and AI refined the execution, showed no improvement over human work alone. The researchers attributed this to output constraints: AI is most effective when it is not working inside someone else’s creative framework. That finding cuts in multiple directions. It confirms AI has creative capacity in specific contexts.

But even so, the strategic intent, the brand’s specific logic and the decision about what the work is trying to accomplish still have to come from people who train AI. When the training comes from creatives who understand what the brand stands for, the system works. When it is absent, AI optimises for engagement without reference to meaning.

Two different games, played at the same time

AI, when trained well by humans, can excel at the short game: maximising response, reducing production friction and optimising against a defined signal. Human judgment excels at the long game: building coherence, earning trust and creating the kind of meaning that makes people return without being prompted. The error is not using AI for the short game. The error is assuming the short game is all there is and failing to ensure that AI reflects a brand’s values. Click-through rates do not help if the ad grabs attention but tells the wrong story about a brand, which is why people are needed to train AI.

For CMOs and design leaders, the practical implication is this: broaden the definition of effectiveness before deploying AI at scale. If the only signal feeding the system is CTR, the system will optimise for CTR. Add brand health indicators to the measurement set and make sure they carry real weight in creative decisions. Protect the strategic choices that cannot be optimised into existence, like the brand’s voice and its values. And use AI upstream, to stress-test ideas and surface inconsistencies, rather than downstream, to manufacture volume.

Clicks are a measure of attention captured. Brand equity is a measure of trust accumulated. Timeless brands know the difference, and they do not confuse one for the other.

Featured in Design Week.