Machines need humans more than humans need machines

The more powerful technology becomes, the more valuable the human perspective becomes. The job of marketing isn’t just to process information, it’s to understand people, writes Goodstuff’s executive director of creative media.

Every so often, a life-changing new technology arrives, and we collectively lose our shit. Steam engines. The motor car. Television. The internet. Now, this one.

In marketing, media and advertising, there’s a ubiquitous narrative that machines are soon to run the whole show. Algorithms will plan campaigns, optimise media, generate creative, write copy, and even brief themselves. Humans, apparently, will stand politely to one side while the robots handle things.

It’s a beguiling story. It’s pretty unnerving to many. It’s also daft.

Because the truth is far simpler: machines may be extraordinary tools, but the entire magical marketing process still begins and ends with humans.

And that matters more than we’re currently inclined to think.

Let’s start with first principles: Marketing begins with human behaviour. Data, after all, is just a structured memory of what people have done. Someone searched for a product, walked into a store, lingered on an online video, and bought some trainers at 11:43 pm

These are not abstract signals. They are fragments of human life translated into numbers.

The process ends with humans, too.

A person sees an ad (maybe). A person interprets it (hopefully). A person has an association encoded in their mind (eventually).

Machines can optimise the delivery of that message with astonishing efficiency, but they cannot replace the fundamental reality that communication is a human-to-human activity, mediated by technology.

Forget that, and marketing becomes a technical exercise instead of a behavioural one.

Tour a car plant, and you’ll see robots everywhere, welding, lifting, assembling with perfect precision.

Machines do the heavy lifting because they are extraordinarily good at repetition, speed, and accuracy. But you’ll also see humans at every critical juncture.

• Researchers identify a need
• Designers shape a prototype
• Engineers design the system
• Technicians calibrate the machinery
• Quality inspectors check the output
• Supervisors intervene when something unexpected happens
• Drivers transport the car to its eventual owner.

No one would build a production line and then remove humans entirely from the process. Not because machines aren’t capable, but because systems drift. Sensors fail. Assumptions break. Context changes.

Humans provide judgment.

Marketing automation should work the same way

Machines handle the scale and precision; humans provide the sense checks that stop things from going quietly wrong.

When AI runs without human intervention, predictable things happen. Biases in training data become embedded in decision-making, algorithms optimise for the wrong metrics, systems hallucinate confidence where certainty doesn’t exist, and automation mistakes correlation for causation.

None of this is malicious (though that’s a near-future risk). It’s simply what machines do.

AI is a probabilistic system trained on past patterns. It doesn’t understand culture, nuance, humour, irony, or human contradiction in the way people do; it approximates them.

Left unchecked, optimisation engines can create marketing that is technically perfect but emotionally hollow; advertising that performs for the machine rather than the audience.

Which is a bit like designing a car purely to satisfy the factory robots assembling it.

I think, nay fervently hope, we’re currently living through what Gartner analysts famously call the “Peak of Inflated Expectations” in its Hype Cycle model. Generative AI has already been placed at that peak in several Gartner analyses, currently only suiting billionaire investors and axe-wielding bean counters.

I’m not suggesting AI is a fad – far from it. But we’re in the noisy phase where people overestimate what the technology can do today, while underestimating the role humans will continue to play.

In other words, we may need a little course correction.

None of this is an argument against AI

When used properly, machine intelligence is a gift to marketing.

It can analyse vast datasets humans could never process, optimise media delivery in milliseconds, identify patterns invisible to manual analysis, and automate the dull, repetitive work that used to eat up days.

Machines are brilliant at scale, speed, and precision. They are useless at empathy, context, and judgment. Which is why the most effective model is not machine-led marketing, but human-guided automation.

Machines do the lifting. Humans steer the direction. We are the “ghosts in the machine” that James Denton-Clark eloquently wrote about.

The irony of the AI era is this: the more powerful the technology becomes, the more valuable the human perspective. Because the job of marketing isn’t just to process information, it’s to understand people.

People are irrational, emotional, contradictory, and cultural. They buy things for reasons that don’t fit neatly into data models. And that messy human truth is exactly why humans need to remain involved throughout the marketing process – from interpreting the signals at the beginning to judging the creative and strategic output at the end.

Machines can help us do extraordinary things, but they still need us to remind them who everything is for. Humans.

Precisely why machines, however sophisticated they become, will always need humans a little more than humans need machines.

Simeon Adams is the executive director of creative media at Goodstuff.