This article was originally published by The Drum.

In CX and marketing, creativity and connection thrive on human elements that AI removes. Graham Wilkinson, Acxiom global head of AI, EVP chief innovation officer, asks whether our drive for seamless automation overlooks the true value of human friction?

For hundreds of years, the goal of innovation has been to make people’s lives easier. To remove friction from processes so we can do things more quickly, easily, efficiently, or at a larger scale. Now, with the advent of AI, we’ve reached a tipping point where innovation is beginning to remove people, and any friction that comes with them, from processes altogether.

Senior leaders acknowledged this in their responses to our 2026 CX Trends survey. About half (49%) think customer experience in their sector will be mostly AI-driven within 10 years. They foresee people only stepping in for very specific tasks. A further 25% think CX will be predominantly AI-led, with essentially no human involvement at all.

So, is this reduction of human contribution a fully positive trend? As we hurtle toward a world full of AI automation, are we in danger of removing human friction from systems that actually feed on that friction? And should we stop to question whether AI is the right choice for every situation, especially when it comes to marketing?

When human imperfection is actually perfect

When business leaders make the case for AI, they often contrast its strengths with humanity’s weaknesses and imperfections. AI agents don’t get sick. AI agents don’t take vacations. AI agents don’t (usually) make silly mistakes (at least not if you set them up right). And yes, there are a lot of situations, such as 24/7 manufacturing, where these non-human attributes make a wholly positive difference.

However, there are also times when the things we see as human weaknesses – things such as emotion, unpredictability, or even naivety – can be real strengths. Imagine a trainee who joins your business fresh out of college. What might be perceived as inexperience or even ignorance can also be seen in the light of inquisitiveness and fresh thinking. For every 10 ‘silly’ questions they ask, there’s sure to be one that makes you sit up and think, and maybe even change the way you do things. AI may be able to complete a trainee’s tasks, but can it bring that same valuable naivety?

It’s easy for humans to overlook the incredible feats our own brains can perform, precisely because our brains are performing them day in and day out. Our multi-dimensional intelligence is informed by our instincts and our wider lives, not just the information classified as relevant to the task at hand. We can assimilate experiences and use our combined senses in ways we can’t even articulate, and which AI, in its current form, can’t replicate. Our hard-wired behaviors, such as the need to eat and sleep, inevitably shape our thinking, which can be both a curse and a blessing. It generates imperfection at times, but it will always make us better at understanding and communicating with other human beings.

In many cases, the human friction we bring, from emotion to distraction, isn’t something that makes us bad at our jobs. It’s precisely what makes us good at them. In those instances, there may well be a danger in replacing real human intelligence with something that’s still essentially just a mathematical model.

Marketing needs chaotic thinking

Marketing and advertising are areas where human imperfection really comes into its own. They’re disciplines where chaotic thinking and human creativity generate the most effective and engaging ideas. Unlike an assembly line at a car plant, it’s not always the same parts, in the same order, going into campaigns. When we turn marketing into a fully automated production line, we end up with a continual reinforcement loop that delivers the mean, not the remarkable. Whether it’s AI-generated marketing content or agent-orchestrated campaigns, everything is ultimately funneled into similar messages and assets – AI’s take on ‘perfection.’

But not all people share AI’s definition of perfection. They might choose a brand based on authenticity or originality. They might be influenced by childhood memories, or by other associations that have little to do with optimizable metrics and everything to do with human experience. When they’re buying a car, people are likely to be influenced by brand affinity, paint color, the sound of the engine, and even the way the interior smells. These are the sorts of things human marketers can understand and that inform their work, even if unconsciously.

The most iconic brands, the most successful products, and the most exciting campaigns are often the result of a real person’s chaotic thinking, and that’s something we shouldn’t underestimate or try to erase.

A reminder to choose intentionally

This isn’t a call for marketers to abandon AI. Far from it. As global head of AI at Acxiom, I’m really excited to see how we’re using the technology to elevate the way brands connect with their audiences. AI decisioning can be very powerful when it’s grounded in clean, connected, and ethically sourced data.

Instead, this is simply a reminder that AI isn’t the answer to everything… a reminder that human imperfection isn’t always a negative thing… a reminder that the question of human versus AI should be an intentional choice in any given situation, and that the right solution will often combine the best of the two.

We don’t yet know what the long-term impacts of automating processes using AI will be. We can speculate that any imperfections and biases built into AI systems (or the data that feeds them) may be compounded and could lead to inappropriate discrimination, erosion of brand, and system fragility. Humans have natural mechanisms for system reset and pressure release, so we need to explore how similar mechanisms can be introduced to AI.

I also think it’s worth brands thinking about their responsibility to human beings, both as people and as consumers. The short-term gain from process automation could ultimately lead to large-scale job loss, massive societal and welfare impacts for future generations, and people who neither trust brands nor have the means to pay for their products and services.

Much of the rationale behind AI adoption involves freeing up humans to do ‘other things.’ However, there seems to be little thought put into what these other things might be. Perhaps we can rebrand them as ‘human things’ – tasks that involve communicating, interacting, empathizing, bonding, and questioning, activities that use our innate human tendencies to nurture AI-based systems, ultimately providing a sensible balance between automation and humanity.

Before we get too swept up in the AI hype, it’s worth remembering that some of our human imperfections are actually vital to our success. And that human friction doesn’t always need to be entirely eliminated from a system to improve it.

This article was originally published by The Drum.