It’s inevitable that Artificial Intelligence (AI) will make its way into the copy and content creation world.
After all, words are data that can be parsed, like any other piece of information.
I’ve written elsewhere about how the future of content and copywriting includes AI — that’s not up for debate.
Platforms like GPT-3 are pre-trained on a massive amount of information — some 45 terabytes of text data run through 175 billion parameters.
That’s the entire content of the internet (at the time of training).
What would you expect from a tool like that?
GPT-3 is a powerful technology that’s transforming the way we produce content intended for human consumption.
So the question isn’t if we “can” use GPT-3 and other AI tools to create content.
The question is: “how” and “when” we should use it.
Let’s explore some possibilities and risks of using AI to generate content, and how it could change the way copywriters and content marketers do their work.
The Rewards of AI and GPT-3: Speed and Volume
One of the biggest challenges that any content marketer or media organization will tell you they have in the content game is simply keeping up.
One of the greatest challenges facing any writer, content marketer or media organization is simply keeping up.
The sheer volume of content produced and published online is something close to 2.5 quintillion bytes of data.
This is where AI-driven content platforms offer a tremendous advantage.
They can spin up content virtually on demand.
For content that’s deterministic — fact-based, without editorial opinion — automated content generation platforms are perfect for the job.
In fact, major news organizations like Reuters, AP and The Washington Post have been using automated content creation tools for years.
The Associated Press has been using Automated Insights’ Wordsmith platform to generate upwards of 4,000 data-driven stories per quarter, from financial reports to sports game recaps.
That’s freed up a lot of time for journalists to focus on more human centered, in-depth reporting.
In the marketing world, AI-driven content generation platforms have sprung up to do the same for overwhelmed content marketers.
And the newest kid on the block, GPT-3, is touted as outstripping anything that’s come before, in terms of generating human-like language and text.
GPT-3 performs better on shorter-length text than long form, but that can still take a lot off of a marketer or copywriter’s plate: think product descriptions, social media posts, and headlines and email subject lines. (Animalz ran a series of experiments with GPT-3 and content generation, with some pretty interesting results.)
And then there’s the use case of creating advertising and marketing for more than one audience.
If you’re targeting different buyer personas, you can spin up an infinite number of variations as required, quickly.
For the right kinds of assignments, AI-driven content platforms can eliminate the effort required to churn out some of the smaller copy assets (at least the first drafts of them) that are part of a content marketing strategy — in record time and with sufficient quality.
The Risk of AI and GPT-3: Bias and Lack of Depth
AI content platforms can generate text that sounds like a human wrote it. Unfortunately, that’s true in more ways than one.
In one sense, these AI-driven content platforms can be all too human.
Because they’re trained on content produced by humans, they embed the same biases and prejudices that are present in that material.
But even as undesirable elements (like bias) show up in auto-generated content, there’s also something missing: insight.
Think of insight as the “human element.”
It’s what makes excellent content worth reading.
AI content generators can absolutely produce human-sounding content, and some (like GPT-3) draw on staggering amounts of data to make it sound like the real deal.
But they can’t create anything new.
When we humans sit down to write, we consume a lot of information (research) to flesh out and mold the idea we started with.
We connect dots to create something new, and we extract insights.
Then — crucially, for authoritative content — we back up those insights with evidence and citations.
AI-driven platforms don’t do that.
They predict “what words most likely come next” based on all the content that’s already written.
They aren’t designed to draw inferences and suggest new ideas.
They don’t go beyond what’s already been said or written because they can’t.
Platforms like GPT-3 may be very good at simulating the style and form of a human writer, across all genres, but they can’t draw conclusions about their subject that get the reader to think about it in a new way.
That’s the job of a writer.
The Future of Copy & Content: A Happy Medium With AI Doing the Heavy Lifting
So where does all this leave copywriters and content creators?
Are they doomed to be replaced by machine learning and AI-driven algorithms in the race to churn out more and more copy and content?
Not at all.
Given the pros and the cons of these content generation platforms, there is absolutely a place for them in a writer’s toolbox.
For one thing, they can help with another big headache that plagues pretty much everyone who’s ever had to write on a deadline: writer’s block.
I’m mostly kidding here, but there’s no denying that using an AI platform to generate a rough first draft of your article may help you get a jumpstart on your deliverable.
Or even if you’re just “stuck” somewhere in your writing, you may be able to get your creative juices flowing again by feeding the AI platform a prompt to auto-generate text as a fresh jumping-off point.
AI-driven platforms will also be really useful for content ideation.
As any content marketer knows, brainstorming ideas for your content calendar isn’t always an easy process.
An AI text generator may help here.
You can feed it the titles of your top-performing blog posts, for example, and have it generate more titles for you.
(Pro tip: With GPT-3 you can set a “temperature” — it controls the randomness of the output, so the higher the temperature the more “creative” the output.)
The important thing to remember is that to create insightful content, the auto-generated content that these platforms produce will still require human oversight.
And I’m not just talking about catching inappropriate biases or potential “logic loops” in longer pieces. You should be on the lookout for “sameness,” too.
Does the content sound like every other article out there?
That’s a recipe for getting lost in the noise.
To sum up: I think it’s reasonable for content creators to use AI-generated content as a starting point, but they should put their own stamp on it before they hit “publish.”
AI and GPT-3 Provides More Opportunities than Risks
I believe that AI-driven content platforms are assets rather than threats to the future of copy and content creation.
The possibilities this technology opens up for experimentation alone are as endless as they are exciting.
The key is to understand when (and how) to use them, and where to draw the line between human and machine.