Why I’ll Never Be Replaced By AI

By Todd Riegler, Digital Cheddar

AI can write a headline, build a spreadsheet, and suggest thirty campaign ideas before I’ve finished deciding whether the first one was any good. That’s useful. It also makes the question pretty obvious: Todd, what exactly are we keeping you around for?

Fair question. I own Digital Cheddar, a Michigan marketing agency, and I use AI in agency work. I have no interest in defending busywork just because a human used to charge for it.

But I’m also not clearing out my desk.

The ‘never’ in this headline is a statement about how I intend to work, not a labor-market forecast or a lifetime employment guarantee. AI will change jobs, including mine. My responsibility is to keep earning my place by making good decisions, understanding the business, and standing behind the work.

A faster first draft gives me more room to do that. It doesn’t excuse me from doing it.

The important part might be missing from the prompt

Picture a business asking for more leads. AI can help create ads, improve a landing page, and suggest ways to increase inquiries. Great. Now imagine that business already has more inquiries than its team can handle, but most are for its least profitable service.

Congratulations. We have automated the arrival of a larger problem.

The useful questions come earlier. Which customers actually fit? What can the team deliver? Where does the money come from? What happens after somebody fills out the form?

AI can help investigate those questions when it has reliable information. Somebody still has to recognize what’s missing, get the right people involved, and check whether the answer matches reality.

That’s a big part of the work I value. A brief is a starting point. Sometimes the most helpful thing I can do is question it.

A confident answer still needs checking

One of the odd charms of generative AI is its ability to make a wrong answer look ready for the board meeting. Nice headings. Clear bullets. Absolutely magnificent nonsense.

This is a documented risk. NIST’s Generative AI Profile describes ‘confabulation’: AI systems can confidently produce false or erroneous content. That makes checking claims, sources, and calculations part of using the tool responsibly.

In marketing, the consequences can be painfully ordinary: an invented product feature, an unsupported promise, a report that draws the wrong conclusion from incomplete data.

If I approve that work, I own the decision. ‘The AI said so’ would be a spectacularly weak explanation to a client whose budget or reputation was on the line.

Humans make mistakes too. The standard should be a process that catches them, with a person responsible for what gets published, recommended, or spent.

Somebody has to say, ‘That’s enough beige’

Generating options is increasingly easy. Choosing something worth putting into the world takes attention.

A headline can be grammatically perfect and completely forgettable. A design can look expensive and feel wrong for the business. A campaign can tick every box in the brief while giving customers no reason to care.

AI can contribute creative ideas, and some will be excellent. My taste gets tested in what I keep, what I change, and what I throw away.

Does this sound like this particular company? Is the benefit clear? Would a real customer care, or are we congratulating ourselves for using the word ‘innovative’ again?

I want tools that help me explore more possibilities. I also want the judgment to close twenty-nine tabs and pick a direction.

Relationships include the uncomfortable conversations

A client relationship involves history, expectations, and consequences. It includes the conversation where the requested idea might waste money. It includes explaining uncertainty without hiding behind a dashboard.

Software can help preserve that history, spot patterns, and prepare for those conversations. That’s valuable. Trust also depends on whether the person making a recommendation listens, follows through, and tells the truth when the answer is inconvenient.

I’m willing to put my name on my judgment. If I need to change my mind, I need to explain why.

You can automate a meeting summary. Earning the right to tell someone their favorite idea needs work is a different undertaking.

My job description has to keep moving

None of this gives me permission to coast. If I define my value as producing things a tool can produce faster and cheaper, I should expect some uncomfortable questions.

AI’s capabilities will keep changing. Today’s limitation is a lousy foundation for a career plan.

The International Labour Organization’s 2025 research points to job transformation as the most likely overall impact of generative AI, because many occupations still include tasks requiring human input. That isn’t a promise that every role survives. It is a reason to pay attention to how the work changes.

My approach is straightforward: get clear on the problem, use the tools where they help, check the output, and stay responsible for the result. Spend the time saved learning the business and making better decisions.

That also means being honest about what deserves a fee. Nobody should pay me a premium for preserving an inefficient process as a historical landmark.

I’m keeping the responsibility

I expect AI to take over more tasks I do today. Good. There are plenty I won’t miss.

What I intend to keep building is the combination of judgment, curiosity, taste, relationships, and accountability that makes me worth working with. That value has to show up in the work. A warm handshake and a strong opinion won’t cover for bad results.

So yes, I’ll use AI. I’ll learn it, question it, and use it to get further than I could alone.

The tools can change. My name is still on the work.