That time I was replaced by ChatGPT

Imagine a marketplace where there are only two products. One product is high quality and is sold at a fair price. The other product is low quality but free.

I used to believe that if you found yourself in this marketplace and were producing the high-quality product, it was time to dust off your resume.

With that said, it was shocking, but not a surprise, when one of my best clients – a creative agency – reached out to me a few months ago to tell me that they no longer required my services as a ghostwriter.

My relationship with this client was built on mutual trust and transparency, so they were good enough to explain that:

1. They were sorry;
2. They had decided to bring the writing function in-house;
3. They were going to use AI to craft articles for their clients;
4. They understood that this would result in lower quality work; but,
5. The cost savings was too compelling to ignore; and,
6. They would be able to apply the savings from 4 to manage any quality issues that might emerge.

My client delivered the news at a time when the tech industry was writing the obituary for the creative class, and writers in particular.

Still, that commentary wasn't computing for me.

John Henry Dies At The End

AI, and Large Language Models (LLMs) in particular, are probability-based models that predict the most likely next word in a sentence based on patterns learned from large amounts of text data.

Yes, these models can infer deeper semantic understanding using context, patterns, and can manage basic ambiguity and inference.

Still, LLMs aren't simulations of the human mind or a surrogate for organic creativity.

While generative AI was providing some impressive results (and still is), that output wasn't a replacement for what I produced, or offering the value that I provided.

Of course, a notification that you are no longer needed doesn't take very long when you're a consultant. Because of this lack of time, and my sadness about being let go, I didn't have the presence of mind to raise the shortfalls of generative AI on my call. Instead, I fumbled over arguments to try to talk my client out of this decision.

As I did, I remembered the story of John Henry, the African American freedman who worked as a "steel-driving man"—a man tasked with hammering a metal bit into rock to make holes for explosives to blast railroad tunnels. John Henry ended up beating the machine that was brought in to replace him.

Then I remembered something else about John Henry's story: He died at the end.

Where I Add Value

My niche as a ghostwriter isn't industry-specific, although my background in finance and data analytics tends to bring in a lot of work from private equity firms and AI companies.

My real niche is uncovering relatable, funny, and impactful stories for companies that have esoteric products and services designed for a specialist business audience.

Those are firms where the senior leadership has grown the firm through word of mouth, offering products and services that are the "best X that you've never heard of" of their respective industries.

These firms have never had to make their offerings exciting for a wider audience. They've just focused on quality and relied on satisfied customers to generate new business.

These companies usually bring me in when that strategy starts getting in the way of growth. Then, I'll be charged with speaking with highly technical professionals who have never had to explain the "why" behind their product… usually after another marketing agency has already tried and failed to make this story impactful for a wider audience.

Duplicating Creative Work With AI

So, yes, I'm a writer. But uncovering the insights and stories that will make the product relevant, impactful, important, and interesting to new prospects takes some detective work. Putting those insights and stories into context takes experience. Making them impactful takes creativity and a sense of humor.

"Could AI do all that?" I wondered.

My own experience using AI was that it certainly could be used as a tool to achieve faster, less expensive thought leadership outputs… if you had an experienced ghostwriter at the wheel, who had already taken the time to speak to the internal stakeholders to get a sense of their insights, priorities, and personalities.

Someone like me.

The alternative was that the clients of my client didn't really appreciate the quality that I thought I brought as a writer.

Or, worse, the end-user did appreciate the quality that I provided… but liked lower prices even more.

What Is Quality Worth?

It didn't take too long to have my questions about quality answered: Two months later, my client hired me back at a higher rate.

I didn't have time to enjoy the ego boost. I earned that money turning around large projects for several clients of client that were about to walk out the door.

Apparently, they had lost faith in the value that the agency provided because the thought leadership being produced was absent both business insight, and quality of craft… and that, in turn, was calling the entire service offering into question.

So, happy ending.

Of course, I'm not kidding myself about the future of ghostwriting and thought leadership.

It's likely that AI is going to lower the price that companies are willing to pay for all writing so, eventually, it's going to lower the price that clients are willing to pay for my writing.

It's also possible that AI writing quality will continue to inch up, as the high volume of low-quality AI-in circulation causes buyer expectations to inch down. If those two trends meet in the middle, high-quality, human ghostwriting will occupy the same rarified space as artisanal bread baking. (And I will be forced to move to Brooklyn and grow a handlebar moustache.)

In the meantime, members of the creative class should not give up the ship just yet.

AI can be a major productivity tool in the hands of experienced writers and other creative professionals.

It does not yet have the capabilities to feel or think—the crucial points of differentiation for creative professionals, even in a world where AI does not exist.