LET'S GET STRATEGIC
AI for Content Creation
by Linda Pophal
A Facebook headline reads, “How to Generate Blog Posts in 1 Minute Using AI.” It’s a sponsored post from Jasper (more about this later), and while the concept is certainly intriguing and tempting—especially when facing a mountain of blog posts to create for various clients—there’s a certain amount of skepticism that goes along with it. Is the hype too good to be true?
Jasper (formerly Jarvis) is one of a number of newly emerging tools designed to help content creators be more efficient in their work—to the point, as the Facebook post suggests, of actually creating content for them. As Payal Dhar explains in an article for The Freelance Creative, the AI engine GPT-3, which is used by both Jasper and Rytr, is built around 175 billion parameters. Fueled by machine learning, which is a component of AI, GPT-3 makes calculations about which words to string together to address a specific need.
Writer is another example. May Habib, Writer’s co-founder and CEO, says, “Marketers need to write lots of content throughout the entire user journey, and there are a lot of parts of the writing process that aren’t especially skilled, satisfying, or fun—like iterating endlessly on blog post titles, rephrasing and summarizing content for social media posts, or creating basic SEO content.” Habib explains, “AI can greatly accelerate the content production process by automating certain kinds of routine, predictable tasks—saving content creators’ time so they can devote more of their time to strategic creative work.”
Dave Rogenmoser, CEO and co-founder of Jasper, agrees. “Content creation tools allow teams to scale their creative operations quickly and with intention,” he says. With these tools, Rogenmoser notes, “Creative teams can write original content faster, boost ad conversions, and engage consumers with unique written content across platforms. Content tools equip teams with rich options that expand their digital footprint, ultimately scaling their media and sales reach.” There is definitely potential here, and the tools are already being widely used in some settings.
THE POTENTIAL AND THE BENEFITS
“Content creation tools can be a time-saver to get answers to threads started, blog posts templated, and key points from long articles summarized,” says Kavita Ganesan, author of The Business Case for AI: A Leader’s Guide to AI Strategies, Best Practices & Real-World Applications and founder of Opinosis Analytics. AI can be used to accelerate the content production process, Habib concurs—automating routine and predictable tasks allows content creators to focus on doing more creative and strategic work. She predicts that in the next few years, more companies “will adopt AI to automate routine tasks, like creating utilitarian SEO content, iterating on hundreds of ad copy variants, or enforcing adherence to style guides—especially as marketers will continue to be expected to do more with less.”
The more that has been written on a topic, the greater the potential of AI tools, says Hannah von Rothkirch, content and social media specialist with Konstruct Digital. Through her own experiences, von Rothkirch says that “today’s AI content creation tools do an OK job of producing content for saturated industries where there is already a lot of content previously written on the topics you want to write about.” Still, she adds, “the content isn’t amazing, and it will never perfectly reflect your brand voice—but it can be helpful to use AI content as a starting point to save time. If you’re lucky, it’ll maybe get you 50% there.”
When it comes to more specialized, niche content, though, von Rothkirch doesn’t think AI will ever be able to deliver. “AI simply cannot produce accurate content if it doesn’t have enough information available to learn from,” she says. But, Habib states, “The next generation of AI content generation empowers content creators to train their own language models, so that AI can better meet their specific needs.” Still, there are some other drawbacks that content creators should be aware of.
THE DRAWBACKS AND THE DANGERS
There are some potential dangers associated with the use of AI tools for creating content. “As a person who understands how these tools are developed, one of my biggest fears of AI content generation tools is plagiarism and misinformation,” Ganesan says. “Things like creating templates [and getting] bullets started and articles summarized are harmless and can significantly boost productivity. But asking these AI tools to generate large blocks of content is the function of what the underlying system has learned,” she says. If the data being fed to the tools contains misinformation, the tools will perpetuate it, Ganesan warns. “There’s also a high chance of stating the words of others verbatim while appearing to be completely original because it’s machine-generated.”
Accuracy of the content created is, or should be, of concern. Writing for the Associated Press, Matt O’Brien shares input from Teven Le Scao, a research engineer working with AI, who says that while these tools are “very good at writing text with the proficiency of human beings,” they’re not so good at being factual. “It looks very coherent. It’s almost true. But it’s often wrong,” Le Scao says.
The pitfalls can become even more nefarious. Hammond Pearce, a research assistant professor at New York University’s Tandon School of Engineering and an expert on vulnerabilities in AI-based systems, says, “We have seen warnings in academic publications as well as informal blog posts that the text-generating models such as GPT-3 can be tailored to produce abusive and distressing content at a frightening pace.” Pearce calls these threats “an automated cyberbully.” Even more concerning, he says, is that they can be used to change the tone of existing text. Despite these dangers, as new use cases emerge and technology becomes more sophisticated, the adoption of tools such as Jasper, Rytr, and Writer is likely. And there are some practical applications for their use.
Ganesan suggests that AI tools are “best used for initial planning, content design, and supporting citation research.” The actual content creation, she says, should be left to humans. Christopher P. Willis, CMO and chief pipeline officer at Acrolinx, agrees. “The key takeaway to this is that you can’t just buy a product and have it create content for you. Language isn’t generic. Point of view isn’t generic,” he states. “There still needs to be a human part of the creative process, and they’ll need to build from what you give them.”
The promise of “turn it on and you’re going to get thousands of words written” is a big goal that’s not likely to be delivered on. You may get a lot of content, but, says Willis, “you’re going to get thousands of words written that aren’t very interesting, that are repetitive and, in a lot of cases, are poorly written.” Moving forward, “Content creation technology will become more sophisticated and intentional. The goal is to remove technological traction in the content space and scale business goals as a result. Our team is pursuing smart browsers integrations and other ways to make the writing experience as seamless as possible,” Rogenmoser states.