The Marketing You Do Best Is What You'll Have to Unlearn First

Perhaps artificial intelligence doesn't threaten the least prepared marketing professional. Maybe it threatens precisely those who have become excellent at what they do.

This provocation appears in Silvio Meira's most recent article and has stayed with me since I finished reading it: what we do best may be precisely what we will have the most difficulty giving up.

It seems contradictory. We spend years studying, working, testing, making mistakes, and honing certain skills. We build careers around them. We develop what we generally call experience. Then a technology emerges that can profoundly change how part of that work is done.

What do we do with all the knowledge we've accumulated?

In the article, Meira talks about a "creative destruction of the cognitive dimension." And he starts from an uncomfortable point: there is a before and after this transformation, and the transition from one to the other will not necessarily be quick, cheap, or simple.

The problem is even greater for those who are already very good at what they do.

This reflection connects to an important idea about learning and change: change doesn't simply mean acquiring new knowledge. Often, it means being able to abandon behaviours, references, and mental models that have worked extremely well for years.

I've been thinking a lot about how this applies to marketing.

Because this may be exactly the moment we are living in.

We're Putting an Electric Motor in the Old Factory

One of the most interesting stories Meira presents is the transition from steam-powered factories to electricity.

When electric motors began replacing steam engines, it would have been reasonable to assume that industrial productivity would immediately increase.

That is not what happened.

For a long time, factories removed the old power source and installed electric motors in essentially the same place. The technology had changed, but the factory's architecture still followed the old logic.

Industrial facilities had been designed around the limitations of steam power. A large central engine mechanically transmitted power to different parts of the factory. This influenced building design, equipment placement, and even how work was organized.

Electric motors eliminated many of those constraints.

But to reach their full potential, replacing one machine with another was not enough. The factory itself had to be redesigned.

When that happened, horizontal layouts emerged, production lines were organized according to workflow, and new ways of distributing machinery and workers became possible. Productivity finally improved.

Electricity was never the problem. The problem was continuing to think as though we were still powered by steam. I see something very similar happening in marketing with artificial intelligence. We are using AI to write a post in five minutes instead of 30.

To create ten versions of an ad instead of three.

To summarize research.

Create images.

Prepare presentations.

Adapt copy.

Generate reports.

Automate emails.

Produce briefs.

All of this is useful. I use artificial intelligence daily in different areas of my work.

For example, I use AI for simpler things, such as daily use of tools to help me with campaign idea suggestions, reviewing texts, and generating alternative titles for social media, to using agents to perform some of the repetitive tasks in my day-to-day work and the operation of my company. These practical applications have changed the way I approach many routine tasks.

But we need to ask a more difficult question:

What if we are installing electric motors inside a factory that was designed to run on steam?

We are producing faster, but we continue to organize marketing departments in essentially the same way. We maintain similar structures, campaigns, customer journeys, approval processes, and, in many cases, even the same metrics.

AI comes in at the end to accelerate a task.

That is efficiency.

Transformation is something else.

Transformation happens when we start asking why that process exists, why that task needs to be performed that way, why that campaign has that particular architecture, and even why we need to produce certain content in the first place.

A campaign no longer needs to begin with a brief, move through creative, media, and approval, and then wait weeks for results before being optimized.

Research, creation, personalization, distribution, experimentation, and learning can become an almost continuous system.

The customer journey could evolve from a relatively static sequence of touchpoints into an adaptive experience.

And perhaps the marketing department of the future will not simply be today's department with a few AI subscriptions added.

It may need to be redesigned.

Some Companies Have Already Realized This

An interesting example comes from the advertising industry itself.

The One Show recognized Monks as its first AI Pioneer Organization. What is most interesting, however, is not the award. It is why the company received that recognition.

The company did not treat AI simply as a way to automate creative production. It created internal training programs, a consulting structure for agent-based transformation, and an engineering unit dedicated to building AI solutions. It also began positioning itself as an orchestration partner, combining talent, technology, and creativity across the content production chain.

In a project with Headspace, for example, AI agents generated and tested creative concepts. More than 460 assets were produced for 20 use cases, reducing production time by two-thirds. According to Monks, the campaign achieved a 62% higher conversion rate.

The part I find most relevant is not the 460 assets.

Producing 460 pieces instead of 46 does not necessarily represent innovation.

What matters is the change in the system: producing, testing, learning, and adjusting begin to work differently.

That is the electric motor beginning to change the design of the factory.

But There Is Another Trap: Being Too Good at the Old Model

This may be the most uncomfortable part of Meira's reflection: the more experience we have, the more we may have to unlearn.

People who have worked in marketing for many years develop something extremely valuable: intuition. A creative director looks at an ad and senses that something is wrong before being able to explain exactly what it is. A copywriter reads a headline and feels that it does not work. A strategist quickly recognizes when a positioning strategy does not make sense. A media professional notices something unusual in the numbers before finishing the analysis.

After thousands of hours of work, some of this knowledge becomes unconscious. We no longer need to rationally reconstruct the entire path that led us to a conclusion. We know. This ability is an enormous professional advantage because it allows us to make decisions quickly, recognize patterns, and avoid mistakes that might go unnoticed by someone with less experience.

But it can also become a trap.

Our intuition was trained within a particular environment, under particular constraints. It was built at a time when writing was expensive, producing an asset took time, creating a video required significant resources, and generating 30 variations of a campaign might have been economically impractical. Many of the rules we learned throughout our careers were reasonable responses to that context.

Artificial intelligence changes those constraints. When producing alternatives becomes faster and cheaper, when testing different versions becomes possible on a much larger scale, and when tasks that once consumed hours can be completed in minutes, some of the rules we learned stop making sense. Not because they were necessarily wrong, but because they were created for a different world.

That does not mean experience has lost its value. I believe exactly the opposite: in an environment where anyone can quickly generate professional-looking copy, images, analyses, and strategies, experience and judgement become even more important. The problem arises when experience stops being a foundation for interpreting what is new and becomes a justification for preserving what is old.

Experience without the ability to reassess can become sophisticated resistance. Experienced professionals need to learn to do something particularly difficult: question precisely what their own careers have taught them was right.

The Jagged Frontier of Artificial Intelligence

There is another problem. It is relatively easy to imagine AI developing in a straight line, first mastering simple tasks, then intermediate ones, and finally complex ones. But that is not how it works. Researchers such as Ethan Mollick describe a kind of "jagged frontier" of AI capabilities, where a model can perform a surprisingly sophisticated task and, minutes later, make an almost absurd mistake on something seemingly simple. Anyone who works with these tools every day has probably seen this happen: you ask for a complex analysis and receive an excellent result, but when you request a minor correction, the AI changes exactly what it was supposed to preserve.

It can produce an extremely convincing marketing plan, complete with segmentation, positioning, channels, KPIs, and a timeline. Everything looks professional, well structured, and ready to present. There is only one problem: the strategy may be wrong. This is one of the biggest traps of generative AI: fluency is not the same as competence.

And that brings us back to marketing. Imagine a company delegating a large portion of its content creation to AI. The copy is grammatically correct, the images look professional, the posts are properly structured, the emails have subject lines, CTAs, and personalization, and the content calendar is full. We have never produced so much. But someone still needs to ask a very basic question: is it any good?

Better yet: is it relevant? Does it differentiate the brand? Does it solve a real customer problem? Does it contribute to business objectives? Or are we simply producing more because production has become cheap?

This is the paradox many marketing departments will face in the years ahead. When production becomes abundant and inexpensive, the real challenge is no longer filling the content calendar. It is deciding what deserves people's attention.

When producing becomes cheap, discernment becomes expensive.

The Risk of Surface-Level Competence

We are entering an era in which it will be extremely easy to appear competent. Someone with little experience can produce an impeccable presentation on almost any subject, generate a SWOT analysis, create personas, develop a content strategy, prepare a campaign, build an editorial calendar, or write a report filled with recommendations. Everything will look professional: structure, appropriate language, a convincing presentation, and a confident tone. But it won't always have depth.

I call this surface-level competence. There is form, but not always understanding; fluency, but not always knowledge; confidence, but not always judgement. And marketing without depth quickly becomes a commodity. If everyone uses the same models to generate the same structures, based on the same statistical patterns, we will end up with an extraordinary amount of content that is technically acceptable and strategically forgettable.

In fact, we are already seeing some of this. Spend a few minutes scrolling through social media, and you will find posts with similar structures, the same phrases, the same hooks, the same lists, and the same conclusions.

Everything is correct, optimized, and perfectly forgettable.

In that environment, competitive advantage no longer lies simply in the ability to produce. It lies in the ability to choose: choosing what deserves to be done, what needs to be questioned, what should be explored more deeply, and what should never have been produced in the first place.

The Marketer Stops Operating and Starts Orchestrating

Meira proposes another idea that I find particularly important: thinking about contemporary work through the combination of three forms of intelligence — individual intelligence, social intelligence, and artificial intelligence.

For a long time, we treated technology as a tool that needed to be operated. You use software, configure a platform, create a campaign, consult a dashboard, and interpret the results. With artificial intelligence, particularly agent-based systems, that relationship is changing. Technology stops being merely something that executes an instruction and starts participating in the cognitive process, helping us research, compare, propose, test, adapt, and even make decisions within defined boundaries.

That changes the professional's role.

The marketing manager of the future will spend less time individually operating dozens of tools and more time orchestrating systems that combine people with different skills, data, institutional knowledge, cultural context, technology, AI models, specialized agents, clients, and customers. The advantage lies in knowing what to delegate to each of these elements and, most importantly, when not to delegate.

That ability requires something no prompt engineering course can solve on its own: metacognition.

You need to know what you know, recognize what you don't, understand where AI is better, identify where it may be wrong, and have enough expertise to audit the result. Receiving a well-written answer is not enough. You need to assess whether it makes sense, is based on sound assumptions, considers the business context, and whether you can defend it in front of a client, a team, or the customer.

This Is Where Experience Becomes Extremely Valuable Again

A recurring narrative is that younger professionals have a natural advantage in this transformation because they adopt new tools more quickly. To some extent, that is true. A young professional can learn a new AI platform in a matter of days and develop considerable operational skill. That has value, particularly in an environment where tools change rapidly, and the ability to experiment is becoming increasingly important.

But knowing how to use a tool and knowing what to do with it are very different competencies. AI dramatically reduces the cost of execution, but it does not eliminate the need for judgement. And judgement is built through knowledge, study, experience, mistakes, client interactions, customer observation, and an understanding of the business.

That is why I believe experienced professionals continue to have an enormous advantage — as long as they do not confuse experience with permission to keep doing everything the way it has always been done.

There is an important difference between having 20 years of experience and repeating the same year 20 times.

The professional who combines experience with curiosity will have extraordinary value because they can do something technology still cannot do reliably: put answers into context.

This is also why the academic foundations of marketing are becoming more important, not less, right now. Technologies change, platforms disappear, algorithms are modified, and tools are replaced. The fundamentals, however, remain remarkably resilient: segmentation, positioning, value proposition, consumer behaviour, brand building, pricing, distribution, experience, relationships, measurement, and value exchange.

Kotler does not need to be discarded because ChatGPT arrived. We need to understand how those fundamentals operate in an environment where execution capacity has changed radically. This is an important distinction between knowledge and tools: a tool increases our capabilities, but knowledge helps us decide where to use it, for what purpose, and within what limits.

So, What Do We Need to Unlearn?

I do not believe the answer is to throw away everything we have learned. Unlearning does not mean forgetting, rejecting experience, or abandoning the fundamentals. It means no longer treating an old solution as a universal truth and recognizing that what worked in one context may not work the same way in another.

We need to unlearn the idea that producing more content necessarily means doing more marketing. We need to unlearn the assumption that campaigns must follow exactly the same processes they followed five years ago, that productivity is measured simply by the volume of deliverables, that specialization means protecting a particular task from automation, and that knowing how to operate a tool represents a sustainable competitive advantage.

We also need to unlearn something even deeper: the idea that our professional value is directly related to the amount of work we can produce. If AI can produce in minutes what once took us hours, insisting on competing solely on execution is probably a losing battle.

But there is another, far more interesting battle. It involves deciding what deserves to be produced, understanding who it is for, defining why it should exist, connecting the initiative to business objectives, questioning the answers, identifying mistakes, recognizing opportunities, adding context, making choices, and taking responsibility for them.

That is still marketing.

The Challenge Is Not Learning AI. It Is Redefining the Work.

Gregory Bateson proposed different levels of learning. At the deepest level, what changes is not simply the answer we give to a problem; the very structures through which we understand the problem change. It is a particularly useful idea for thinking about the moment we are living through.

Learning to use ChatGPT, Claude, Gemini, or any other platform is important. Learning prompt engineering can improve results, building agents can increase productivity, and automating processes can reduce costs. But all of this may still represent learning within the existing model.

Transformation begins when we ask a different question: if we were building our marketing department today, from scratch, with all this technology available, would we organize it the way it is organized now?

Would it have the same roles, the same processes, the same suppliers, the same meetings, the same campaign cycles, the same KPIs, and the same deliverables? I find it difficult to believe the answer would be yes. That is probably the point.

Artificial intelligence is not simply giving marketing new tools. It is changing some of the economic assumptions on which we built our work. That is why learning will not be enough. We will have to unlearn.

And, ironically, the professionals who may benefit most from this transformation could be precisely those with the greatest experience — provided they have enough courage to question some of the things that made them experts in the first place. Operating tools will become increasingly easy, production will become increasingly inexpensive, and generating alternatives will become almost instantaneous. But discernment, empathy, context, knowledge, responsibility, and judgement will remain scarce resources.

In the end, the most important skill for a marketer in the age of artificial intelligence will not be knowing how to produce the answer. It will be knowing how to recognize which answer is actually worth using.

The marketing we do best today was built inside a factory designed for steam.

The electric motor has arrived.

Now we need the courage to redesign the factory.

References

(May 15, 2025). Monks Is Named First AI Pioneer Organization by The One Show. Monks. https://www.monks.com/articles/monks-and-named-first-pioneer-organization-in-one-show

(2024). Personalized GenAI Creative Case | Headspace | Monks. Monks. https://www.monks.com/case-studies/headspace-personalized-at-scale-performance-creative

Dell’Acqua, F., III, E. M., Mollick, E., Lifshitz, H., Kellogg, K. C., Rajendran, S., Krayer, L., Candelon, F. & Lakhani, K. R. (2026). Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality. Organization Science. https://doi.org/10.1287/orsc.2025.21838

Meira, S., Neves, A., Belfort, R., Calegario, F. & Garcia, V. (2023). O que você sabe sobre as Inteligências individual, social e artificial?. Proximo Nivel. https://proximonivel.claro.com.br/inteligencias-artificial-social-humana/

Tosey, P. & Mathison, J. (2008). Do Organizations Learn? Some Implications for HRD of Bateson's Levels of Learning. Human Resource Development International 7(1), pp. 105-118. https://doi.org/10.1177/1534484307312524