Talk Commerce Talk Commerce
Back to Blog
Brent Peterson
Blog
| 8 min read

Will AI Replace Human Writers?

By Brent W. Peterson

My Conversation with Christian Brim on The Profitable Creative

I had the opportunity to sit down with Christian Brim of CORE Group on his podcast, The Profitable Creative, to talk about something that matters to every creator, business owner, and entrepreneur right now. The episode, titled “Will AI Replace Human Writers?” digs into the tension between AI-generated content and human-created work, and where the real opportunity lives for people willing to lean into both.

The Content Cucumber Origin Story

Content Cucumber has been around since 2018. A group of college entrepreneurs started it as a human-first writing service, and my previous company Wagento was one of their earliest clients. I bought the company in April 2025 after spending years in the e-commerce space, including running Wagento, and founding Content Basis, an AI-forward content creation company.

Content Basis launched right around the time ChatGPT arrived. It was AI-forward with a human editor in the loop. What I discovered over three years of running it was revealing. Clients loved the idea of getting content fast. They loved the price point. But most of them did not want to spend time reading, editing, or refining the AI output. They wanted a magic box. When they realized that the magic box still required their involvement, engagement dropped.

Content Cucumber represents the other side of that coin. It is built on the foundation that human writers create better, more authentic content. My goal now is to find efficiencies within that human-first model, add strategic services like SEO and content planning, and use AI to support the humans doing the writing rather than replace them.

Why AI Did Not Kill Content. It Exposed Bad Content.

One of the points Christian and I kept returning to is this idea that AI did not destroy the content industry. It revealed just how much low-quality content already existed. When everyone suddenly had access to tools that could generate a thousand words in thirty seconds, the flood of mediocre content became impossible to ignore.

Audiences noticed. People are learning to spot AI-generated writing. There are tells. Overused words like “delve.” Formulaic structures. A voice that sounds polished but says nothing original. When every company’s blog reads the same way, people stop trusting any of it.

I have attended dozens of e-commerce conferences over the past two years, and this theme keeps coming up. Vendors and brands are discovering that their AI-generated content sounds identical to their competitors. If your brand voice is indistinguishable from every other company using the same LLM with similar prompts, you have a differentiation problem that no amount of automation can fix.

The Editing Use Case is Where AI Shines

Christian and I found real alignment when we talked about using AI as an editor. He shared that his most productive use of LLMs in his accounting and writing work is as a review tool, something that goes beyond spell-check and grammar into consistency of thought, writing style, and voice alignment.

That mirrors my approach exactly. I write something, then I apply what I call my “brand” prompt. I have a large document that captures my writing style, my preferences, terms I avoid, and structural guidelines. The LLM checks my draft against all of that and flags inconsistencies.

I take it further. I ask the LLM to count adverbs in a document and tell me where each one is necessary or unnecessary. LLMs overuse adverbs by default, and most human writers do too. Getting a grid that shows every adverb with a recommendation to keep or cut saves real time in the editing process.

This is the role where AI earns its place. Not as the creator, but as the quality control layer that helps human writers produce cleaner, more consistent work faster.

RequestDesk.ai and the Brand Voice Problem

During the conversation, I talked about RequestDesk.ai, a tool I started building in 2024 to solve a specific problem. Most businesses struggle with brand voice consistency. When you have multiple writers, freelancers, agencies, and AI tools all producing content, the output sounds like it was written by a committee. Because it was.

RequestDesk stores brand identity in a structured database. When content gets generated or edited, the system injects brand context before the prompt executes. It also checks output against a database of overused AI terms and flags anything that does not match the brand’s established voice.

The idea is simple. Whether a human or a machine writes the first draft, the final output needs to sound like the brand. Not like ChatGPT. Not like Claude. Not like a generic corporate voice. Like the actual business it represents.

The LLM Consistency Problem

Christian asked a sharp question during our conversation about the unpredictability of LLM outputs. He noted that in accounting and coding, getting correct answers only 80% of the time is not acceptable. I agreed, and I shared a real example.

We do product injections into Shopify blog posts. The code that dynamically generates and injects product references works correctly about 80% of the time. The other 20%, it does something wrong. Or as I like to joke, it works right 90% of the time and wrong 20% of the time.

This is the fundamental challenge with LLMs. They are not algorithms. An algorithm produces consistent, predictable results. An LLM produces variable results. Some days it performs brilliantly. Other days, the same prompt with the same context produces something completely off. That variability is exactly why human oversight is non-negotiable. You need scaffolding, templates, rules, and checkpoints to catch the 20% that goes sideways.

Content Strategy, Not Just Content Volume

One of the shifts I am pushing at Content Cucumber is moving beyond just writing blog posts to providing real content strategy. When I reviewed years of customer feedback, a pattern emerged. Clients loved the writing quality, but they consistently wished for more strategic guidance around SEO and content planning.

The difference matters. Writing a well-crafted blog post is valuable. Writing a well-crafted blog post that targets a specific keyword gap, addresses a product that currently appears on page 100 of Google results, and includes proper structured data, JSON-LD, and Q&A sections for search discoverability is transformational.

The typical Shopify client has fewer than a hundred words on a product page. That was never enough for good SEO performance, and it is even less sufficient in a world where both traditional search engines and AI-powered answer engines need rich, structured content to surface products and brands.

The flywheel subscription model we built at Content Cucumber addresses this directly. It includes a monthly interview with the client to capture their voice and priorities, blog post creation, newsletter writing, social media content, and ongoing strategic adjustments based on Google Search Console data. The human interaction component is the most important part.

Humans Are Still Having Better Ideas

This might be the most important takeaway from the entire conversation. After years of working with AI tools across content, coding, design, and strategy, I keep arriving at the same conclusion. Humans are still generating better original ideas than LLMs.

An LLM will give you dozens of ideas. It will brainstorm endlessly. But the ideas tend to cluster around the obvious. They reflect patterns from training data. A human strategist who knows a specific industry, understands a client’s competitive landscape, and has spent years building pattern recognition through lived experience will consistently surface ideas that an LLM would never generate.

The best workflow I have found pairs human ideation with AI execution support. Humans decide what to write and why. AI helps with research, editing, formatting, and distribution efficiency. The human stays in the driver’s seat at every decision point. That is not a philosophical preference. It is a practical conclusion based on what produces the best outcomes for clients.

Where We Go From Here

The conversation with Christian reinforced something I think about constantly. The future of content creation is not human versus AI. It is human with AI. The creators and businesses that figure out how to blend authentic human voice with AI-powered efficiency will outperform everyone trying to automate their way to relevance.

AI did not replace human writers. It gave the best human writers better tools. It also created a massive opportunity for businesses willing to invest in authentic, strategically sound content at a time when the internet is drowning in AI slop.

That is the bet I am making with Content Cucumber, with RequestDesk.ai, and with everything I build going forward.
Humans always in the loop.
Not sometimes.
Not when it is convenient.
Always.

You can listen to the full episode on The Profitable Creative.
Tag @ChristianBrim and use #theprofitablecreative
Follow The Profitable Creative on Instagram @profitable.creative
CORE Group at @coregroupus

This has been produced in cooperation with Content Cucumber
https://www.contentcucumber.com/

Follow Talk Commerce on your favorite platform:

Tags

AI content creation AI slop brand voice Brent Peterson Christian Brim Content Cucumber content strategy e-commerce content human writers human-first content LLM editing podcast recap RequestDesk SEO content The Profitable Creative

Author

Brent W. Peterson

Brent W. Peterson

View all posts by Brent W. Peterson
Back to all posts