THE CYBORG GUIDE TO AI COPYWRITING

Part #1: Why AI Has Changed Copywriting (Permanently)

Click below to skip straight to the desired section:

1. The 3 Things Everyone Gets Wrong About AI
2. Why AI Has Permanently Changed Copywriting
3. Example: How I Found Product/Market Fit
4. Six Copywriting Applications Where AI Shines
5. The Future Of AI In Copywriting & Marketing
6. Next Part: How To Write Copy With AI

Or simply continue reading below…

The 3 Things Everyone Gets Wrong About AI Copywriting

Within just three months of its launch, ChatGPT reached 100 million users and became a household name.

By comparison, I’ve been explaining what “copywriting” is to my friends and family for the last 11 years, and they still ask me, “What do you do again?” whenever the topic comes up.

The recent AI revolution has been a wildly misunderstood mixture of generation-defining technology, exaggerated claims, and outright lies.

And as a result, I need to explain why everything you’ve read about AI copywriting up to this point is probably wrong.

First, no AI tool has come anywhere close to the “1-click” Holy Grail.

AI as a “1-click solution” to virtually anything is a wild goose chase propagated by Venture Capital.

The best AI tools are able to automate around 80% of writing workflows that previously had no real automation options, but that last 20% supplied by the human user — usually focused on strategy/direction, editing/course-correction, and “extra mile” content additions — are the difference between creating something that is legitimately profitable and something that is completely worthless.

The mainstream rhetoric around AI, both heralding it as a 1-click writing savior and bemoaning its failure as such, are a red herring to this entire conversation.

Second, AI writing’s primary value is connected to opportunity cost.

The fundamental mechanism of today’s AI writing tools (built on Large Language Models, which I’ll explain later) results in a regression towards the mean.

In other words, these tools are powerful, because they can take a large data set of human writing and create new writing that exists right down the middle of everything in that data set. 

• What’s the most common argument made on a topic? 
• What are the most common ideas associated with it?
• What is the most common language used?

These are the things that AI is designed to give you.

When businesses attempt to build workflows around this type of content — usually with the goal of simple, blind scale at a minimal budget — it’s essentially worthless. What value is it to a brand to have the average of everyone else’s copy and content?

On the other hand, when businesses and freelancers attempt to plug this tool into a HUMAN writing process, it eliminates hours of low-impact work, freeing up that time for human writers to make their content exceptional and steer AI’s potential in support of real marketing objectives.

Third, AI language models are nearly as good as they will ever be.

There’s been a common sentiment in the market that we are approaching AGI (artificial general intelligence) territory, and that the ability of GPT4 and other LLMs to produce writing that sounds human is indicative of this approach and only the beginning of a new revolution.

Anyone who has spent even a small amount of time looking into AGI knows this sentiment is nonsense — the result of tech-bro, twitter-influencer pseudoscience — and has absolutely no basis in reality.

As we just talked about, Large Language Models are designed to give us the average of what humans have already written, and in my opinion, GPT4 is already SO good at this goal, that there’s only marginal room for improvement. In fact, we are MORE likely to see a decline in quality from here, as the companies who own these models reduce their abilities for the purpose of safety/liability AND low quality, purely AI-generated content is cycled back into model data sets.

There’s a potential solution to these two issues, but we’ll address that solution later in this piece.

TLDR: Entrepreneurs and businesses seeking to use AI as a cheap, one-click marketing solution are in for a bad time, both now and in the future. 

Why AI Has Permanently Changed Copywriting

When most people think about writing, they think about creative self-expression.

Historically, writing has been humanity’s #1 tool for chronicling ideas and sending those ideas across space and time.

With this context in mind, it’s easy to understand a view toward AI as an irrelevant, unhelpful, or even antagonistic player in the writing space.

But when it comes to copywriting specifically, self-expression is irrelevant and creativity is often counterproductive. Copywriting is a business tool, and based on the way we approach business in the 21st century, the ONLY thing that matters is profit.

So what makes writing drive profit for a business?

1. Attracting attention to a business.
2. Converting attention into revenue

That’s essentially the complete list.

When we dig deeper into attracting attention, there are a lot of different things we can try, but here’s the aspect that is most important to our discussion.

Increasing the quantity of efforts we make to attract attention and the variety of things we try directly results in an increase in revenue.

This isn’t a secret. It’s common knowledge. When we do more to attract attention, we make more money.

We see something similar when converting attention into revenue.

Conversion is a function of finding market fit; this can be content/market fit, product/market fit, service/market fit, etc.

In order to find that fit initially, most businesses have to experiment with different messaging and offer angles. The same product sold to the same people might do incredibly well when communicated one way and completely bomb when communicated a different way.

Once again, increasing quantity and variety in conversion messaging directly results in an increase in revenue.

And that continues to be true even after the initial fit is established.

The way people respond to specific messaging evolves over time and market conditions. Products and offers change. The market changes. The more quantity and variety of messaging a business can experiment with, the more chances they have to maintain or increase revenue.

So why don’t businesses infinitely scale that quantity and variety?

The answer is cost. The job of most marketing and sales teams (outside the context of massive funding rounds) is NOT to increase revenue at any cost. Their job is ultimately to increase net profit, so scaling attention at a positive ROI is important.

Want to increase attention? Want to try more conversion angles?

That’s gonna cost you, and historically speaking, it cost a lot.

To get an experienced copywriter to help you with all this, you’re looking at around:

• $500 per blog post
• $200 per email
• $5,000 per sales page
• $200 per ad
• $1,000 per web page

And that’s very middle-of-the-road pricing. There are no guarantees at these price points.

And more importantly, there are no true guarantees at any price point. Business is an ongoing experiment, and every new copywriting deliverable is always a hypothesis.

Within this context, we start to understand why a tool that can massively increase quantity and variety at virtually no cost… changes everything.

Let’s look at a real-life example to illustrate this in action.

Example: How I Used AI To Find Product/Market Fit

Two months back, I was approached by an entrepreneur with a supplement business built around his personal story of overcoming alcohol addiction (now 9 years sober) via the aid of supplementation. He used a Facebook advertising funnel to attract and convert customers to his business, and his product/market fit was built around helping people quit drinking.

The problem was that the FDA had reached out to him and told him he was no longer allowed to say anything about alcohol addiction or helping people quit drinking.

Now what?

We basically needed to find an entirely new product/market fit — a way to sell this product using a new message that didn’t include any of the product’s main selling points.

Here’s how this process would have worked 5 years ago.

1. We collaborate on a new messaging angle to try.
2. He pays me $10,000.
3. I write new ads, a new landing page, and a new VSL script.
4. We run at least $1,000 in ad spend through the funnel.
5. We evaluate results.

If we’re really lucky, we’ll make back more than $5,000 in revenue on this funnel, and everyone will be happy.

The much more likely scenario is that we don’t hit profitability on our first swing, and that means we have a lot more work to do.

Do we try to improve the existing funnel? That’s another $2,500 to make a whole new round of ads and a few more versions of this landing page.

Do we start over from scratch with a new angle? That’s another $7,500 to create a whole new funnel with a completely new messaging angle.

And we still are at risk of striking out again and starting over from scratch.

If the business doesn’t have at least $20,000 to sink into this process, what do they do? Well, they can either look for an entirely different way to drive sales, or they can go find a much less experienced copywriter who can do all this for $5,000.

The odds of success drop significantly along with the price point.

That’s how it used to work.

Here’s how it worked for my client thanks to AI.

1. We agreed on a $10,000 price point for successfully finding a new product/market fit.
2. We collaborated on three different messaging angles to try.
3. I used my AI-assisted writing process to rapidly create funnels for all three angles.
4. We tested all of them simultaneously.
5. One was profitable.

This is the power of AI when it’s used the right way.

We had the same goal we’ve always had: profit. And we used the same strategic approach that’s always worked to find that profitability.

The only difference was time and cost.

AI’s efficiency allowed me to spend a lot less time on this project and by extension, charge less to the client, while still providing them with the strategic value of 11 years of copywriting experience.

When speed and cost matter, AI is an undeniable game-changer.

And when do speed and cost NOT matter?

Six Copywriting Applications Where AI Really Shines

Everything we’ve covered up to this point basically boils down to: if you use AI the right way, it can help you write 3-4 times faster.

But what does “the right way” actually look like?

At this point, there’s six main writing applications where AI really outdoes itself:

1. Outline and content structure
2. One-line content
3. Summarizing basic information
4. Examples and analogies
5. Emotional language
6. Proofreading and editing

Let’s take a closer look at each one.

1. Outlining and Content Structure

GPT4 in particular is really good at structuring content. By contrast, humans are really, really bad at structure… especially humans who have chosen writing as their profession.

One of the reasons I’ve had so much success with SEO copywriting is that I’m able to structure content like a good teacher — creating context and then walking people from A to B to C in a way that makes sense and keeps them engaged.

AI can do this too. It doesn’t have the core understanding of marketing objectives necessary to do this at a top-1% level, but its algorithmic approach to content is good enough to create outlines better than 90% of the writers I’ve worked with over my 11-year career.

2. One-Line Content

There are many types of one-line content used in business, including:

• Headlines
• Taglines
• One-line ad copy
• Social media captions
• Short product descriptions
• Email subject lines
• CTAs
• Etc.

AI can instantly create massive quantities of these, and then a business can test all of them or have a human select the ones they like best.

3. Summarizing Basic Information

There are many different types of content. A blog post like the one you are reading now revolves around my personal point of view and can only receive limited assistance via AI.

A blog post explaining what ChatGPT is, on the other hand, is going to include a lot of generally-available information that needs to be summarized for the reader.

For example, let’s say I want to explain what a Large Language Model is to my reader.

I could just write out my personal definition, but that’s going to be a boring process and my definition is likely to be biased towards my arbitrary experiences around the topic. What most people do is go look at other definitions, but then you have the problem of trying to rephrase or paraphrase without plagiarizing.

AI allows us to generate an “original” description of any topic that we like, create multiple versions, and pick and choose from what we like best.

4. Examples and Analogies

When teaching a subject, few things help cement a concept as well as examples and analogies. These devices help our brain connect new ideas to familiar mental models.

Using more examples and analogies in my content is something I’ve wanted to do for years, but time has been a limiting factor. Coming up with an example or analogy that is effective at illustrating the concept I want to teach often took up upwards of 30 minutes.

GPT4, on the other hand, can create exceptional examples and analogies in seconds.

5. Emotional Language

A surprising problem that many copywriters (especially those outside the direct response world) deal with is that we are too deep into the marketing bubble to remember how normal people perceive copy.

When we see dramatic storytelling or highly emotional language, we often roll our eyes and think, “Nobody would actually fall for this.”

But that’s the problem… they do fall for it.

Visceral, jarringly emotional language is one of the most effective tools in the direct response toolkit. The more you can lean into the emotion of what someone is feeling, even to a degree that would make a telenovela jealous, the better you can resonate with their pain and connect their desires to your offer.

This is an area where GPT4 is sneakily strong. When you ask it to “lean into the emotion”, it will often go massively overboard, giving you a never-ending string of dramatic adjectives, phrases, and analogies.

If you are approaching ChatGPT here to give you one-click copywriting, this output is completely worthless.

If you are approaching it as a collaborator, on the other hand, not only do you now have a large list of legitimately-strong phrases to choose from in building your narrative, but you are also going to feel more comfortable writing telenova-level copy after reading this over-dramatic nonsense.

6. Proofreading and Editing

GPT4 is a great proofreader and a reasonably useful editor.

It’s not going to be perfect, and I’ve actually started catching typos out of ChatGPT over the last month, but I would rather use ChatGPT as my proofreader for three distinct reasons:

• It’s free.
• It’s around 90% perfect, and the last 10% does not impact revenue.
• It’s instant.

That last one is especially important for anyone with a rapid publishing schedule.

It’s convenient having GPT4 do my proofreading for me, but I can get this same functionality from a large number of other tools.

What I can’t get from other software is a decent developmental editor, and that’s where GPT4 offers something new.

It will consistently provide reasonably strong suggestions for improving a piece of content.

What About Research?

You may have noticed that I left out anything related to research, which is probably the main function writers are using GPT4 for today.

I’ve tried to limit this summary to writing-specific applications, because covering all the non-writing applications that can be used by writers would require a massive series of articles.

But more importantly, research and data analysis connect to where all this AI technology is taking the market, and I want to discuss that in more detail below.

The Future Of AI In Copywriting And Marketing

Beyond copywriting, there are a lot of other things that AI is being used for right now:

• Personalization at scale
• Research
• Brainstorming
• All sorts of micro-automations

And while each of these (and many more) are significant in their own ways, ultimately, none of them represent where AI is taking the market.

Data access and analysis is the true revolution that GPT-3 began.

In 10 years, the way this tech has changed how we interface with the internet or segmented batches of information will be 100x more impactful than how we use it as a writing tool, and I’m saying that as someone who is adamant that it’s a writing game-changer.

What if, instead of condensing our search queries into 4-5 word phrases, we could very clearly spell out exactly what we are looking for online and have an AI find it immediately?

Google has been making anti-consumer and anti-publisher choices for the last decade.

It’s trash, but it’s the best trash we’ve had up to this point.

OpenAI, Google, and Facebook are all working toward this right now. None of them have made much headway… yet. But it’s just a matter of time.

In 5 years, the old search engine model will be completely obsolete.

In another vein, while ChatGPT is useful already, it’s limited by its need to source from its entire training data set. As solely GPT-4 generated content makes up a bigger and bigger portion of content on the web, broad data sets will be increasingly worthless.

In the future, people will use a hybrid model where large data sets are used for certain aspects of training, while the AI can be made to give larger weight to specific subsets of data for specific purposes.

Imagine being able to train an AI model to replicate a specific writer or brand voice.

Imagine having an AI-search assistant that can comb through every podcast, youtube video, or article from a specific source and answer your hyper-specific questions similarly to if you asked the source author.

This is where AI is headed. It’s the potential solution to the brand safety and regurgitated-AI content problems I mentioned at the beginning of this article. And in terms of the applications I’ve mentioned here, I think it’s a good thing.

We live in an information age where most people are unable to access that information efficiently, and while this won’t fully solve that problem or the disinformation problem closely connected to it, there’s a chance it could help.

My goal is to help individual service providers and entrepreneurs better understand and leverage this technology, because it’s here, we can’t roll it back, and the larger players in the market WILL leverage it to exploit us every chance they get.