AI, Research, and the Death of Intuition
Data tells you what happened. Only intuition tells you what to do about it.
In the summer of 1998, I was working on a campaign for a British retailer that was losing market share to a more fashionable competitor. We had data. We had mountains of data. We had focus group transcripts and brand tracking studies and competitive analyses and customer satisfaction surveys. The data told us everything: what the customers thought, what they wanted, what they feared, what they valued. The data was comprehensive, reliable, and completely useless, because the data could not tell us what to do. It could describe the problem with exquisite precision, but it could not generate the solution. The solution came from a conversation I overheard in one of the retailer's stores — a customer explaining to her friend why she kept coming back despite the dowdier image. That conversation contained, in a single overheard phrase, the insight that became the campaign. The data had missed it entirely.
I tell this story not because data is worthless — it is not — but because it illustrates a principle that I believe is under mortal threat: the principle that intuition, informed by experience and observation, is essential to the creative process, and that no amount of data can replace it. This principle has been under attack for decades, as the advertising industry has become increasingly data-driven. But the arrival of AI-powered research and analysis tools has accelerated the attack to the point where I fear the principle may be lost entirely.
What AI Research Tools Do Well
I want to begin with what works, because I am not an ideologue and I do not believe that fairness is optional. AI research tools are extraordinarily good at processing large volumes of information. They can analyse thousands of customer reviews in minutes. They can identify patterns in social media conversations that would take human researchers weeks to find. They can synthesise competitive intelligence from multiple sources and present it in clear, actionable summaries. They can do all of this faster, cheaper, and more consistently than any human research team.
These are genuine capabilities, and they have genuine value. A creative team that has access to comprehensive, well-organised research is better equipped than a creative team that does not. I have no quarrel with the tools themselves. My quarrel is with the way they are being used, and the way they are reshaping the relationship between research and creativity.
The Insight Problem
The fundamental problem is this: AI can identify patterns, but it cannot generate insights. I am aware that this distinction sounds pedantic, but it is not. A pattern is a regularity in the data — "customers aged 25-34 prefer product variant B" or "negative sentiment increases during the winter months." A pattern is a fact. An insight is an interpretation of that fact — an understanding of why the pattern exists and what it means for the brand. Insights require empathy, imagination, and contextual understanding. They require the ability to see beyond the data to the human behaviour that generated the data. They require, in short, intuition.
When I overheard that customer in the retailer's store, the data already contained the information that a significant minority of customers remained loyal despite the brand's declining image. This was a pattern, and it was clearly visible in the numbers. What the data could not tell me was why they remained loyal, and it was the why that mattered. The customer's offhand remark — which I will not reproduce here, because even decades later I feel a professional obligation to protect the client's confidence — revealed a motivation that no survey question had been designed to uncover. She was not loyal despite the brand's unfashionable image. She was loyal because of it. The lack of pretension was not a weakness. It was the brand's deepest strength. This insight transformed the campaign, and it could not have been generated by any analytical tool, no matter how sophisticated, because the analytical tool would have classified the customer's loyalty as a data point rather than a human story.
The Erosion of Judgement
What concerns me most about the current enthusiasm for AI-powered research is not the tools themselves but the epistemological shift they are accelerating. There is a growing belief in our industry — a belief that is rarely stated explicitly but that shapes decisions at every level — that data-derived conclusions are inherently more reliable than intuition-derived conclusions. This belief is seductive because it sounds scientific. It sounds rigorous. It sounds like progress. But it is, I believe, profoundly wrong, at least as applied to the creative process.
The creative process is not a scientific process. It does not proceed from hypothesis to experiment to conclusion. It proceeds from immersion to understanding to leap. The immersion phase involves absorbing as much information as possible — about the product, the consumer, the market, the culture. The understanding phase involves synthesising that information into a coherent picture. And the leap phase involves jumping from that picture to an idea that is not contained in the data, that could not be derived from the data, that surprises even the person who generates it. This leap is the creative act, and it is, by nature, intuitive. It cannot be validated in advance. It can only be recognised after the fact.
AI research tools are superb at the immersion phase. They can provide more information, more quickly, more comprehensively, than any previous technology. They are adequate at the understanding phase, provided the understanding required is analytical rather than empathetic. But they are useless at the leap phase, because the leap phase is not analytical. It is imaginative. It is the moment when the creative practitioner sees something that the data does not show — a connection, a tension, a possibility that exists only in the space between the facts. This is where intuition lives, and this is what is being eroded.
The Bernbach Parallel
Bernbach fought this battle in the 1950s, long before AI was even a concept. The enemy then was not artificial intelligence but market research — focus groups, copy testing, concept validation. The tools were different, but the dynamic was identical. The researchers argued that creative decisions should be grounded in data. Bernbach argued that data could inform creative decisions but could not make them. "We are so busy measuring public opinion that we forget we can mould it," he said. "We are so busy listening to statistics that we forget we can create them."
This is not anti-intellectualism. Bernbach was a deeply thoughtful man who spent enormous amounts of time understanding consumers. But he understood that understanding is not the same as obeying. The creative practitioner must understand what the consumer thinks and feels and wants. But the creative practitioner must then go beyond that understanding to show the consumer something they did not know they wanted — to reframe the product, the category, the conversation in a way that changes the consumer's perception rather than merely reflecting it.
AI cannot do this. AI can reflect the consumer's existing perceptions with unprecedented accuracy. It can map the landscape of opinion with extraordinary precision. But it cannot change that landscape, because changing the landscape requires an act of imagination that is, by definition, not contained in the existing data. The existing data describes the world as it is. The creative act proposes the world as it could be. These are fundamentally different operations, and conflating them — treating the descriptive as though it were the prescriptive — is the great epistemological error of data-driven marketing.
What Must Be Preserved
I am not calling for the abolition of AI research tools. That would be foolish, and I have enough foolishness in my life without adding more. The tools are useful, and they will become more useful. What I am calling for is the preservation of intuition as a legitimate and essential component of the creative process. Intuition is not guesswork. It is not mysticism. It is the product of years of experience, observation, and practice — the accumulated judgement of a practitioner who has spent decades learning to read human behaviour, to sense cultural currents, to identify the unspoken needs and desires that data cannot capture.
The agencies that thrive in the AI era will be the ones that use data and intuition together — that use AI tools to deepen their understanding and then trust their creative people to make the leap from understanding to idea. The agencies that replace intuition with data — that treat the AI's output as the answer rather than the starting point — will produce work that is comprehensive, well-informed, and creatively dead.
Data tells you what happened. Only intuition can tell you what to do about it. Protect the intuition. Everything depends on it.