What AI Still Needs From Us

Sep 21

As AI takes on more of marketing execution, the quality of the human input matters more than ever.

We hear “human in the loop” all the time when we talk about AI.

I use the phrase too.

As the founder of an AI-forward marketing agency, I believe strongly that humans need to remain involved as we integrate AI into more of our work.

But I've been thinking about what that actually means.

Having a human somewhere in the process isn't automatically enough. The bigger question is: What is that human contributing?

For marketers, I think that's where this conversation gets much more interesting.

Marketing has always been deeply human

At its core, marketing is about understanding people.

It's noticing why someone responds to one message and ignores another. It's understanding the context around a customer problem. It's recognizing a cultural shift before there's a report explaining it. It's taking something you've experienced, observed, heard in a conversation, or learned from a customer and connecting it to a business problem in a way that wasn't obvious before.

That's the part of marketing I've always loved.

We're creative problem solvers.

And as AI gets better at doing more of the execution around us, I think that ability becomes even more important.

Consumers are already telling us that authenticity matters in AI-generated experiences. Accenture found that 41% of consumers distrust AI-generated content when it lacks authenticity, while 45% distrust it when it lacks personality.

There's a reason for that.

People still want to feel like there's someone on the other side who understands them.

AI can orchestrate more. Humans need to give it something worth orchestrating.

This is where I've started thinking differently about the role of AI in a marketing organization.

We're moving beyond asking ChatGPT to write a caption.

We can build agents that research, analyze information, generate content, execute tasks, connect systems, and help orchestrate workflows across the marketing function.

That's an enormous opportunity.

At Social-Kraft, we're constantly thinking about where AI can give marketers more capacity and help teams work better. The more I learn about what's possible, the more excited I get about where this is going.

But more automation also makes the quality of the input incredibly important.

An agent can have access to your brand guidelines, customer data, content library, campaign history, processes, and every piece of context you've documented.

Someone still has to give it perspective and bring the idea that didn't already exist in the system.

That's the human contribution I don't want us to lose.

Authenticity has to enter the system somewhere

I think sometimes we treat authenticity as the final quality-control step.

AI creates something. A human reviews it. We make it “sound more human.” Then it goes out into the world.

I believe that's too late.

The human perspective should be present much earlier.

Give AI the customer conversation you couldn't stop thinking about.

Give it your opinion before asking for its opinion.

Give it the observation you made walking through an airport, sitting in a meeting, talking to your sales team, or spending time with someone completely outside your industry.

That's very different from opening a blank prompt and asking AI what your brand should say today.

The first approach uses AI to amplify human thinking. The second risks allowing AI to become the source of the thinking.

And when millions of marketers are working with many of the same models, the difference matters.

The marketer becomes the orchestrator

I don't think the future marketer needs to personally execute every part of the marketing process.

Actually, I think the opposite is happening.

We're going to orchestrate more.

We'll have AI systems and agents helping us research, create, personalize, analyze, optimize, and execute. The marketer's job will increasingly involve deciding what should happen, what context the system needs, where human judgment matters, and whether the output actually serves the customer and the business.

That requires a different kind of skill.

Taste matters. Curiosity matters. Understanding people matters. Business judgment matters.

The technology can give us leverage on all of those things, but it needs something meaningful to leverage in the first place.

Human in the loop needs human in the input

I'm very optimistic about what AI is going to make possible for marketing teams.

I want us to automate more. I want to build agents. I want marketers spending less time on repetitive execution and more time solving interesting problems.

But I also think we have to protect the part of ourselves that makes those systems valuable.

Human in the loop shouldn't mean putting a person at the end of an automated process to approve what the machine created.

It should mean intentionally designing human perspective, judgment, creativity, and experience into the system from the beginning.

Because the goal isn't to make AI sound human.

It's to give AI something genuinely human to work with.

Karina Jimenez