Artificial Intelligence has been around for quite some time, but the introduction of ChatGPT in 2022 has changed how we see and use AI in copywriting and content marketing workflows. But there are still many questions left to answer:

Will it eliminate jobs?

Can it destroy the delicate SEO and content marketing ecosystem?

Will Google get all ticked off and update its algorithm again to penalize AI-generated content?

And most importantly: Can a chat-based AI really create content that holds up against human creativity and nuance?

Experimenting With Artificial Intelligence and Content Writing

We thought we’d take a shot at it. We opened up ChatGPT, came up with a quick test concept, and started plugging input through the chat interface. And after about 45 minutes, we took what we had generated and matched it up to a basic website wireframe and layout concept.

And this is how it all went down:

Starting Out

We started with an imaginary client looking for solutions to the most common pain points for grant writers. Our first prompt was detailed, which helped set us up with some excellent content to work with from moment one.

Pain points were easy for ChatGPT to generate. For artificial intelligence and content marketing, this is a fantastic first step that lends itself to the strengths of having a large body of knowledge available at a moment’s notice. The AI was trained for just this kind of input.

Now let’s take this list and make it more “human.” We’re going to change our input strategy to a more conversational request:

Refining Our Results

This works great, and it’s perfect for copy or social media starting points. Let’s mold it a bit…

Now we have a brilliant block of copy for a simple FAQ section. But something is missing. Since we don’t actually have this imaginary SAAS product, we don’t have the final S in the P-A-S framework: Solution. Let’s see what we can do about that:

Alright, so asking for software solutions did provide an answer that could easily be made into a blog post with almost no human interaction. But if we want something for our specific software, we need to give ChatGPT something better to chew on. (Also of note: it looks like we hit a limit for a singular output. We’ll condense from here on to make sure we get complete outputs.)

Okay, you showoff. That was an accidental submission. And yet, our robot buddy did something we were planning to do by hand. Moving forward.

Getting More Detailed

Now we’ve coordinated pain points with solutions specific to our imaginary software. We’re officially selling it now. Let’s switch back to FAQs so we can have a nice, clean, and directed page to work with.

Hmmm…what if we wanted a landing page instead? Let’s kick it up a notch. (FYI: our right-hand machine had a hiccup here. We had to resubmit the input and wait quite a while to get a response. And by quite a while, I mean a minute or so.)

Ah, there we go. This first output is close, but we refined it a couple of times before calling it quits.

Hitting Its Limits

And it looks like, by the last output there, we’ve hit a roadblock. Up to this point, ChatGPT has been running like a well-oiled…machine, I guess. But we are now at the point where a human touch may be necessary. Still, we had fun, didn’t we?

Let’s do one final input before we move on to filling out our wireframe:

Good work, little buddy. Good work.

The Results of Our Artificial Intelligence and Content Writing Experiment

It’s easy to end here and get smug about how not clever those last couple of outputs were. But let’s not forget all our little robot pal has accomplished up to this final point:

An excellent list of pain points for our target audience
Some great conversational copy for a natural-sounding campaign on social media
A FAQ that could be nearly copy-pasted into a blog post that points to our imaginary software solution
Imaginary features for our imaginary SaaS solution, plus a sales pitch
Another product-specific FAQ for our company site
P-A-S framework copy for a landing page
Long-form P-A-S content for a website
Some awesome springboards for a clever landing page, funnel, or ad campaign (with just a few tweaks to supercharge it)

And, when we included the time it took to jot down our responses through the process, it all happened in about 45 minutes.

How Do Artificial Intelligence and Content Writing Work Well Together?

Overall, it’s smart to combine artificial intelligence and content writing in most contexts. There are so many things we can get stuck on when working on our own. It’s easy to become lost in title modification, and trying to force out one more version of that old CTA out of your brain could make you go mad. AI helps push through this, and it helps move you on to better things.

Less Time Spent on the Basics

Outlines. Lists. Features. Pain points. All these things take a ton of time to write up, and they often feel bland or dull. Tools like ChatGPT or Jasper.ai make it possible to hand those basics over to someone else. And AI tools do this kind of work exceedingly well. They present new angles, offer fresh wording, and spit it out in seconds when you might have spent hours on the same thing.

Less Time on Information Discovery

Research is an integral part of good writing. But collecting data for a new subject area can feel overwhelming, especially if it’s a highly technical field. B2B businesses require a greater depth of knowledge to write targeted copy. And putting artificial intelligence and content writing together in a single step means tapping a vast knowledge base in an instant. But as you’ll see below, checks and balances are necessary to make this work for you instead of against you.

Less Energy on Legwork and Surface-Level Tasks

Have all the information you need, but it’s unformatted? AI can help with that. Our FAQ lists are proof of this: even if you had to input all the data to get your AI up to speed (which you might, depending on your product or service), it’ll still crank out a simple list faster than you. And it’ll format it automatically.

More Time for In-Depth Processes

Now that you’ve got your basics put together in a matter of seconds, you’re free to spend your hours exploring the deepest depths of your content topics. More research, more advanced comparisons, and better analyses are possible more often. Yeah, you might have the time for occasional deep-dive articles already, but imagine if you could provide that kind of quality content for every article because you’re not slogging through the simple stuff.

More Time on Creative and Format

Social media copy takes time to write. Social media creative elements take time to create. If you’re short on time because you’re running a business or providing marketing for multiple accounts, your use of artificial intelligence and content writing becomes an opportunity to turn your attention on more design elements like infographics and slide decks. And since they get much better engagement, you’re improving every aspect of your social media reach by adding AI to your toolbox.

More Time for Iteration and Fine Tuning

Solid titles are the gateway to your content. Get them right and you’ll see traffic. Get them wrong and you’ll see flat lines in your numbers. AI is a powerful tool to make sure you don’t get them wrong. But even more, AI allows you to come up with a fresh concept that you can then tweak, twist, manipulate, and mold into that perfect title.

What Doesn’t AI Solve for Content Workflow?

Notice something about the positives of combined artificial intelligence and content writing? Never once does AI stand on its own at this point. Even though it can do an exceptional job sometimes, these little machines can’t be left to their own devices.

We had to use multiple inputs for some of the outputs to match our desired vision. And we had to change little details here and there and even change course entirely to pull out high-quality content. ChatGPT is pretty agile (as is Jasper Chat), but AI isn’t ready to take the wheel yet.

It’s Not an Autopilot System

If you’re planning to crank out content at lightning speed, be ready to be disappointed. What does come out typically requires considerable work, and sometimes you’ll have to start over because those well-intentioned robots took a left turn when you needed them to go straight.

Proofreading Is Still Necessary

More obviously, you’ll need to proof for some of the stranger anomalies that pop up with machine writing. Most of the time, you’ll find small things that don’t feel quite right, like the AI isn’t speaking its native language (spoiler: it isn’t). Other times, you’ll see facts or figures that aren’t quite right. And in the worst cases, you’ll find all-out falsehoods that have to be corrected. I don’t want to be too hard on them, but sometimes the robots are just trying to make an educated guess and miss the mark.

User Intent Must Be Added

This is where I saw a bigger weakness. And it has to do with the higher-order thinking skills that actual humans have that our machine assistants don’t yet have at the same intensity. User intent takes a great deal of analysis, evaluation, and discernment to get right. And while AI often has a good idea of what’s going on with a request, it may not have the intent dialed in right away. Our first FAQ serves as a perfect example.

That list was accurate, and it was logically sound. But the AI tried two separate directions before we got our input right. The first attempt was too generic to be useful to people looking for a software solution. The second attempt directed the reader to products that weren’t our own. By the third time, after defining our imaginary software’s key features, we were able to coax out the proper content for our intent.

It Still Requires Outside Expertise

Be mindful of what you let your AI buddy do on its own. It loves to make up stuff, not because it’s trying to make you into a liar, but because that’s how it naturally works. And in specific products, services, or industries, you’ll have to feed plenty of information into the AI to generate something that truly works to describe your company’s offers. Product and industry experts are necessary to get that ideal content.

Basic Wit, Not Pattern Disruption

Incorporating wit into the output was definitely the biggest weakness in this experiment. While we’ve seen solid outputs at other times, ChatGPT just seemed to have lost its panache this time around. Seriously, that final prompt. A cute rhyme is great for a jingle, but it’s not exactly wordplay. Your human experience will easily outperform machine humor any day. And that means that pattern disruption—stopping the scroll or increasing page dwell times—is still a truly human artform at this point in history.

How Will We Combine Artificial Intelligence and Content Writing?

Tomorrow could bring a whole new iteration of the current GPT-3 algorithm, but what we tested today gives us some important information for utilizing artificial intelligence and content writing to save time and better use our energy.

For now, AI is a great writing assistant.

You can offload some of the most repetitive or frustrating parts of writing content and still keep quality close to the same. You can reformat in an instant. You can summarize information quickly. And you can mix and remix your outputs until you get the bones you need to build a solid article or campaign.

But AI is still a co-pilot for now.

We anticipate some changes to our own workflow, but the core of our content creation process will still fall into our hands, not the machines.

We’ll be using AI to:

Destroy writer’s block
Fast-track outlines and lists
Reformat some of our information and research
Ideate and iterate titles, headings, and sales copy
Generate more ideas for posts, blogs, and keywords
Develop stronger strategies based on available information

We will still:

Write long-form content by hand
Do in-depth research to build rich articles
Craft compelling headlines, copy, CTAs, and subject lines
Build out content strategies, but with AI as an assistant to our process

In short, AI can cut down on lower-order thinking tasks so we can enhance everything we do with more experience, expertise, authority, and trustworthiness. For us, combining artificial intelligence and content writing will provide an improvement to our quality, much more than an increase in our quantity. Not only that, but it will offer us the opportunity to focus on content strategy at a deeper level: more data, more research, and more user intent than ever before. And the power of AI makes it possible.

The Final Result: Our (Very Basic) AI-Generated Website Mockup

Now that you’ve gotten through all the sticky stuff, let’s see how our AI-generated content fits into our wireframe:

Every piece of text in this example, save a single title, was written by our AI assistant. And it’s fascinating to see how quickly we could populate a design with fairly compelling copy. It only took about 10 minutes to build this template out with ChatGPT’s original text.

What Works Well

Personally, the FAQ section is the most impressive part. That set of questions and answers provides a strong foundation for customer discovery. But the key features are also impressive. The homepage also offers a decent tagline that catches attention fairly well. For copywriting, ChatGPT did a nice job of meeting the basic expectations for a SaaS landing page.

What Needs Improvement

But it’s not perfect. There’s some serious optimization that could improve the appeal of the product. We tried to tease it out of our little robot, but it just couldn’t make it happen. The best example is on the fourth section. The CTA is generic and doesn’t connect with the pain points strongly enough. More solution than agitation, and it honestly needs a bit more of a poke to get people revved up for the software.

The Final Takeaway

Feel free to incorporate ChatGPT, or any other AI generator, into your workflow. The outputs are generally intuitive, and they feel pretty human in many cases. But we won’t be putting the AI in charge of any final products for the foreseeable future. As a sounding board, it’s top-notch. The only thing better is a room full of creative people and an espresso machine. Take those fresh ideas and start making them your own.

AI will give you the framework you need. It’s up to you to build out a memorable experience with your content.

Machine Mayhem: How a 1-Hour Experiment with Artificial Intelligence and Content Writing Created a Complete SaaS Landing Page