It’s nearly impossible to ignore the buzz surrounding generative AI models, like OpenAI’s ChatGPT, and their potential for content creation. These technologies are undeniably impressive in their ability to generate convincingly human-like text—and can do so quickly. But if you scratch just below the surface, you’ll start to uncover some major limitations when it comes to developing successful, effective editorial content.
You get what you pay for
In their current state, these AI systems lack nuance, critical-thinking skills and the ability to tell original stories and apply strategy and audience considerations. And importantly, they cannot guarantee accuracy. If our goal as marketers is to create content for our clients that’s strategic, audience-centric, accurate and ownable, then the technology falls way short of delivering. Sure, AI chatbots can create content fast—and cheap—but the result will likely be generic, repeatable and, if not inaccurate, unreliable.
We tried, it failed
We’ve been experimenting with generative AI’s capabilities for a few months now but decided to put ChatGPT to the test in a hypothetical assignment: Write an SEO-driven pillar page on infographics for our blog. The result? It shaved two days off the typical writing process. But that’s where the good news ended. No one was happy with the output. Even when given detailed direction on style, tone and audience, the platform created an article that required additional edit rounds of edits and reviews. It contained none of our brand voice and wasn’t suitable to position Imprint as a thought leader. Oh, and because it doesn’t source its information, it couldn’t be fact-checked. Ultimately, we ended up taking just as much, if not more, time to complete the assignment. (So much for faster and cheaper.).
So with all this in mind, Imprint won’t be using generative AI technology to create new content. We also won’t allow our writers to use the technology to develop copy, as we need to guarantee to our clients we’re creating original material that’s sourced and fact-checked. As it stands, we simply can’t trust AI-generated content to be any of those things.
Giving the tech its due
That’s not to say marketers should write off AI completely and forever. Some early users have described the technology as an alternative to Google for its ability to respond to questions. In fact, the AI arms race in internet search is already on. Microsoft announced a huge investment in OpenAI (creators of ChatGPT), and that Bing search will be incorporating the technology. Google scrambled to respond, unveiling its proprietary chatbot technology, Bard, that it will weave into Google search.
While we certainly have to consider the risk with regards to accuracy, these newer AI platforms could possibly be used as brainstorming tools and potentially as an efficient way to compile background information.
AI chat technology is evolving, and as it does so will the ways we’re able to use it. Some future applications could include translation or even hybrid human/AI work models. But for now, content marketers should view generative AI models more as a potential tool than as a content generator.
Where do you stand on generative artificial intelligence?
What has your experience been with these technologies? We’d love to hear your thoughts on the pros and cons, and what you think the next year or two could have in store. Let us know below!