Do you still need us? Or could a robot create engaging, share-worthy content at a much more affordable price?
If I were a B2B marketing manager with one eye on the budget, I’d know which outcome I’d be rooting for. As easy as we are to work with, managing an AI is probably a heck of a lot easier.
So, at the risk of shredding our client contracts, here’s the lowdown on the state of AI in content marketing.
Robots Are Already Writing Some Copy
There is a significant amount of robot-created content on the internet already — some of which is disturbingly good.
Artificial intelligence is used in three main ways to create content at the moment, explains Ravn Research founder Clare McDermott:
- “Formulaic content” that is often data-focused and produced in vast quantities.
- “Creative support” tools that accentuate the ability of writers.
- Chatbots that have private conversations with customers on websites and social media.
“Most people might not think of the second or third methods as genuine forms of content creation, even if they do play a valuable role in marketing teams,” McDermott writes. “But the first use, formulaic or not, is written content by any standards.”
This kind of content is created by a software process called natural language generation (NLG), says futurist Bernard Marr. It can be used to create multiple forms of content, including business reports, emails and social media messages.
Each of these forms has a distinct style and structure, he explains. In some cases, the format will need to be defined to create the right kind of content. But sometimes, AI can do this by itself. There are already several NLG-based tools, Marr notes, including Quill from Narrative Science, Polly from Amazon, and The Washington Post’s in-house tool, Heliograf.
The vast majority of these tools are used to create content from reams of data, writes personal branding and marketing coach Wendy Marx. “For example, The Washington Post has used AI to sift [and] examine sports scores and create articles that summarize local games, as well as post updates to Twitter about those games,” she explains. “The LA Times has been using AI to analyze earthquake data and turn around breaking news articles on earthquakes throughout the world.”
Storyline Labs’ Vasili Shynkarenka uses GPT-3 — an AI program we’ll cover in detail below —to create blog post titles that perform impressively on Hacker News, a social news site. Despite never having more than one “point” (roughly the difference between upvotes and downvotes) on the site before, Shynkarenka’s first AI-powered headline garnered 229 points and 126 comments in a single day. Over the next few weeks, four more AI titles reached the front page resulting in 1054 upvotes and 37,000 site visits.
Far from facing scorn, these tools are welcomed by writers who can’t parse data as quickly as AI can.
Luckily For Us, Other Content Leaves a Lot to Be Desired
Creating highly templated articles that turn complex data into something halfway readable is one thing. But what about articles that explore a question or teach readers how to do something. Can AI write those?
Erm, not so much.
People are trying to create meaningful, creative content with technology like NLG, but the result is the kind of cheap, questionable content loved by unscrupulous agencies. Created for pennies on the dime, this content isn’t meant for human eyes. In fact, you’d have a hard time making sense of this content if you tried.
Natasha Lomas at TechCrunch took one such tool, Articoolo, for a spin back in 2016: “Fans of cryptic crosswords, who relish the mental adrenaline kicks they get from being forced to slalom from cliché to climax, with precious little in the way of linkage, may also be unlikely appreciators of the robot oracle’s semantic gaps. Normal, average readers, however, will struggle to locate much in the way of sense.”
Such content is designed to game Google, which any SEO pro will tell you is doomed to fail. It’s a strategy from the early 2000s, writes Heather Lloyd-Martin, CEO at SuccessWorks, an SEO content training and strategy firm. Back then, when brands could rule Google by spinning content, it didn’t matter whether the copy was terrible and the grammar sloppy. Fortunately for us all, Google’s algorithm has gotten much, much smarter since then.
Fast forward to the present day and AI-created content hasn’t improved all that much. Daniel Priestley, cofounder and CEO at startup accelerator Dent Global, used GPT-3-powered content creator Copy.ai to write a post titled “Can an AI write business blogs?”
It only took the program three seconds to create the content, but the output amounted to 10 one-sentence lines and a short conclusion totaling less than 200 words. Hardly the comprehensive piece Priestley was hoping for.
But he did manage to draw some positives. “On the whole…it’s not a bad starting point for an interesting article and it would be a lot easier for a busy entrepreneur to edit or expand this blog post rather than starting from scratch,” he writes. “In practical terms, you could run the AI system a dozen times in under a minute and pick a reasonable starting point.”
GPT-3, which stands for Generative Pre-trained Transformer, is one of the biggest AI breakthroughs in years. It can learn how to imitate human writing and could even be the first step toward creating artificial general intelligence, says The Verge’s James Vincent. While there’s a lot of potential, it’s hardly the finished article, however.
As Vincent notes: “Close inspection of the program’s outputs reveals errors no human would ever make as well as nonsensical and plain sloppy writing.”
AI Can’t Completely Replace High-Quality Copy Written By Humans
Whether you’re running a newspaper or a B2B blog, nuanced subject matter requires a human writer. That’s because AI has some significant issues, according to John A. Tures, a professor of political science at LaGrange College. In particular, AI doesn’t understand things like satire or mistakes.
It’s a matter of original thought. AI is limited to the information we feed it, says growth marketer Sujan Patel. “To fully replace manual content creation, AI has to be able to think like a human. It has to be able to feel (to have emotions), it needs to form opinions, and it needs to think critically.”
While AI can do some of the leg work, such as suggesting optimized titles or providing prompts, it can’t create wholly original material, says product marketer Sophia Bernazzani. As a result, blog posts and books “will likely be written by humans for the foreseeable future.”
That’s because high-quality, original content is a fundamental part of content marketing. It’s no longer possible to cheat Google with low-quality, keyword-stuffed articles that no one will ever read. Backlinks are what matters. And you can only earn those backlinks if your content is well-written and thought-provoking enough to warrant sharing.
AI Can Be Used in Other Areas of Content Marketing
While you can’t replace writers with robots, you can augment their efforts with AI. There’s a lot more to content marketing than just the writing. Much of these tasks are ripe for automation.
When it comes to specific and repetitive tasks, AI is much more effective and efficient than humans, writes journalist Shane Richmond. “Anything that involves crunching large amounts of data and spotting patterns is a good target,” he says. “It doesn’t forget things and it doesn’t get bored – both of which can be helpful for certain jobs in content marketing.”
Take Market Brew, for example, which uses AI to make intelligent on-page SEO suggestions. By simulating all possible optimizations and comparing their effectiveness, the software is able to make ROI-based recommendations.
Grammarly, the AI-powered spelling and grammar checker, helps writers improve their copy, as Mike Kaput, chief content officer at Marketing Artificial Intelligence Institute, describes. The software doesn’t just highlight errors; it makes intelligent suggestions on ways to improve. Grammarly isn’t perfect, though, so don’t do away with your in-house editor just yet.
AI-powered content marketing tools don’t end there:
- Kontentino can automate all aspects of social media promotion.
- Keywee uses consumer data to target your most relevant audience.
- Jenni identifies gaps in your content and offers suggestions.
- Writesonic helps copywriters create variants of their copy for headlines and landing pages.
Brands and Agencies Should Prepare for an AI-Based Future
It seems like you can’t get rid of us — and there’s a chance you never will.
AI is already used in content marketing operations, and robot writers are churning out full-length articles more quickly than I can write this sentence. But when it comes to high-quality content creation, humans are yet to be matched.
Still, artificial intelligence is here to stay in content marketing. As business coach Lilach Bullock observes, it’s going to play a much more significant role in the future. AI will become a core tool used by marketers at every stage of the buyer journey.
The important takeaway, then, is not whether you can replace writers with AI, but how you can get the best of both worlds. A core responsibility for marketing managers today is striking a balance between human insight and AI effectiveness in the best way possible.
Images by: Franck V, Shahadat Rahman, JESHOOTS.COM
- How to Get the Best Ideas From Your Subject Matter Experts - July 12, 2022
- The SaaS Buyer’s Journey: Discovering New Solutions - June 14, 2022
- 15 High Authority Guest Posting Sites for SaaS Brands - May 31, 2022