Welcome to the future, where vacuums run themselves, cars drive themselves, and content writes itself. Sort of.
Artificial intelligence, machine learning and robotics have brought all kinds of convenience and innovation to our lives, but the human touch is still necessary — as anyone who’s ever had to dislodge a Roomba from under a couch knows. For all the hype around self-driving cars, the National Highway Traffic Safety Administration makes it clear that “every vehicle currently for sale in the United States requires the full attention of the driver at all times for safe operation.” And if you’ve ever fantasized about plugging a couple of sentences into an AI writing assistant and churning out long-form blog content that will drive traffic to your site and convert it into leads while positioning you as a trusted expert in your field … well, you can do the first part pretty easily, but all bets are off when it comes to the rest. AI is great at getting things started, but it won’t get you too far along on your journey without human intervention.
You’ve probably heard about ChatGPT, the AI chatbot made available to the public by OpenAI late last November (and if you haven’t, congratulations on avoiding the internet for the last three months). Research firm OpenAI, which is also behind AI image generator DALL-E, had received $3 billion in investment money from Microsoft prior to the launch of ChatGPT (a widely reported $1 billion in 2019 and a lower-key $2 billion since), and will receive another $10 billion based on the runaway success of ChatGPT. If it was one of the biggest tech stories of 2022, it’s poised to be even bigger in 2023.
The New York Times called ChatGPT “the best artificial intelligence chatbot ever released to the general public.” However, it’s being used as far more than a chatbot. In a time when content marketing is more important than ever, many are turning to it for long-form content like blogs and articles.
Using AI for content creation isn’t new. AI writing assistants, many of which currently use OpenAI’s GPT-3 model, have been around for years (the November 2022 release of ChatGPT marked the rollout of GPT-3.5, so it is a half-step ahead of them; GPT-4 is expected this year). Paid tools like Jasper, Anyword, Texta and Writesonic market themselves as “the future of writing,” able to create “unique and readable content” that “generates and optimizes your copy,” producing “high-quality articles, blog posts, landing pages, Google ads, Facebook ads, emails, product descriptions, and more in seconds.” Sounds like just the ticket for gearing up your content marketing game, right? It can be — within boundaries. There are ways you should use AI writers in content marketing, and ways you shouldn’t.
Do: Generate ideas
If you’ve ever stared at a blank page for hours and called it “the creative process,” you know getting started is more than half the battle. Whether it’s a neatly organized outline or a chunk of semi-sensical word vomit, just having something on the page can help jump start the process. Writing assistants are great for this because if there is one thing AI can do well, it’s spit out a lot of words that are mostly relevant to the topic at hand.
For instance, when beginning this article, I asked ChatGPT to write an outline for a long-form article on content marketing in the age of AI. It created a five-part outline, each part with two to three bullet points, that was certainly usable — just fill in the blanks and you’d have an article. I also asked it to generate an introductory paragraph on the topic. While I didn’t use either to create the final product, having it on the screen helped me feel like I had made progress, which was a definite psychological benefit.
Do: Generate titles
Titles are hard to write. For every 10 submitted pieces of content I receive, two or three come in without titles — writers can churn out 1,500 words of content, but writing a compelling title, one that captures the essence of the piece, gets readers’ attention and is the right length, is hard. AI is helpful here. Whether you feed them your completed article or simply tell them what you’re writing about, you can get most AI bots to spit out several titles, and even if you don’t like any of them as is, chances are there is at least one that will do the job with just a little tweaking. Much as with generating an outline, sometimes you just need a starting point.
Do: Generate social media posts
You’re not creating content marketing pieces for the sole purpose of having them sit on your website waiting for people to find them (although you’re not not doing that either — a lot of the buzz around ChatGPT has centered around its SEO capabilities). You need to drive readers to your site, and social media requires a little creativity too. AI does a decent job here. You can ask ChatGPT to create a Twitter post on your blog’s topic, or input a few paragraphs of text, and it will churn out exactly 280 characters including hashtags (this is important, because one of the hardest parts of writing for Twitter is getting all the information you need into the character limit). It will also turn out longer content for LinkedIn or things like Instagram captions. It’s often been said it’s harder to write short than long, so this is one area where AI can be a big help in your content marketing projects.
There are more ways AI can aid in content marketing, but these are three solid items in the “do” column. Now, onto the don’ts.
Don’t: Use AI for research — or anything you’re not going to fact check
One of the known issues with the GPT-3 learning model is that its knowledge base ends in 2021. In some cases, this may not matter —there’s already a lot of controversy surrounding ChatGPT’s use in academic papers because an analysis of Romeo and Juliet or a paper on the causes of World War I don’t necessarily require updated information — but for subjects that are constantly evolving, like technology and science, it’s very likely you’ll end up with a piece of content that’s outdated at best and factually incorrect at worst. In a very meta example, I asked ChatGPT, which runs on GPT-3.5, released in 2022, to tell me about advancements in GPT. It told me, “The latest version of GPT (Generative Pre-training Transformer) is GPT-3, which was released by OpenAI in June 2020.”
To its credit, ChatGPT is open about its lack of knowledge. While there was no disclaimer in the answer to the previous example, the page itself does have some boilerplate text about its limitations. But when I asked it who won the National League batting title last year, figuring I’d catch it out with “last year,” rather than giving me the answer for 2020, it told me its knowledge base cut off at 2021 and I should check the MLB website. Score one for the robot. However, there’s still plenty of risk, and one of ChatGPT’s most well-known and widely reported flaws is the tendency to output information that is just flat-out wrong, regardless of the time frame. If you’re using it to create content for your company website, presumably you know enough about your topic to catch any incorrect info — just be sure to check the output.
Don’t: Use it to be the voice of your brand
AI writers don’t have a voice. I know, I know – you’ve seen all the hilarious examples of ChatGPT-generated content in the style of Shakespeare or Dr. Dre. Using its vast preprogrammed knowledge, AI can certainly mimic just about any voice you like, but it can’t create something that doesn’t already exist. Building a brand includes building trust with customers, and a unique brand voice is essential to that. Your customers should recognize your brand as easily from your content as they do from your logo, and spitting out AI-generated content and slapping it on your site as is won’t do that.
Don’t: Assume AI is static
With as fast as things move in the world of AI, what you read last week or last month may no longer be accurate. You may have been avoiding AI because you heard Google penalizes AI content. You may have heard AI-generated content violates Google’s spam policies, or that the Helpful Content system means it will damage your SEO. But Google’s algorithms evolve constantly and the AI landscape is developing rapidly, so do your own research, ideally from primary sources (Google, OpenAI, etc.) and don’t depend on knowledge gained from static sources — including this article. Be sure to check the publication (or at least the “updated”) date on any source you read.
Is AI for your content marketing a do or a don’t?
It’s both. Can AI replace humans when it comes to creating basic business content? It would be all too easy (and self-serving) to throw a blanket “no” over that. But the truth is, AI serves an extremely effective purpose when it comes to making content marketing accessible to more businesses at a time when content marketing is more important than ever.
As with all forms of robotics and automation, AI likely does pose a threat to the very lowest-level human writers; those, for example, with no formal training who are writing in their non-native languages for less than a penny per word (go ahead, check Upwork or Fiverr, they’re a dime a dozen — no pun intended). But you’re not going to build your brand with that type of content any more than you are with unedited AI content — unedited being the key word.
AI is as good as the amount of work you’re willing to put into it. That may include feeding it the research links you want to base your blog off of, using it to create outlines that a human then expands upon, or allowing it to create a starting chunk of content that you then fact check extensively and rewrite to have a human voice (and pass AI content detectors, because those are a thing and as AI becomes better, so will they).
There are plenty of uses for AI, and ChatGPT in particular, and not all of them involve writing long-form content. Just go on LinkedIn to see examples, citations and debates on the subject. However, when it comes to the very specific purpose of content marketing, AI has its place — but just like self-driving cars, that place requires a human guardian. For now.
is editorial director of BPO Media’s publications Workflow and The Imaging Channel, and senior analyst for BPO Research. As a professional writer and editor, she has specialized in the office technology industry for the last 20 years. Prior to that she worked in public relations and has a master's degree in communication arts.