AI helped me write this blog post. At least, it helped me write part of it. The text you see below italicized and in purple is a direct copy and paste from a demo of an algorithm I was playing around with on Sudowrite.com—just one of a growing number of text generator platforms in the market today. What this tool and others like it do is learn from existing text (i.e., these words you’re reading now) to generate new text that reflect the characteristics of the original without repetition. So what I’m doing with this blog post is feeding the AI my human-written text and allowing it to produce new text. Let’s see how this goes…
While the AI-generated text is far from perfect and still requires human supervision, it is quite remarkable.
Word-for-word copies of existing text are not that interesting. But the AI can also generate text that reflects style and structure with a lot less hassle.
The AI-generated text below is also a quick and dirty job, but I find it much easier to comprehend than the copy-pasted text in purple. With limited human supervision, the AI has generated text that is familiar with my style and thought process but it is different enough to be interesting.
So the next time you read the news and groan at the inane and repetitive stories, just remember that there will soon be AI that can mitigate a lot of that work.
This is a very early look at a very new technology. The results below won’t be entirely representative of what we’ll see in the future, but they do show the potential and strength of AI to produce high-quality content.
And to be fair, the folks behind Sudowrite don’t position their AI-powered writing assistant as a means to replace human-generated text. But rather, they position it as a way for writers to “bust writer’s block” and spur new ideas for where to take a story next. Analogous tools, such as the one found on Jasper.ai make similar claims, but go a step further to suggest they can write blog posts, website copy and more.
So it makes me wonder, what other implications does AI hold for content marketing?
My first thought was that this could be a new channel for machine learning and AI to filter into the content marketing world. Similarly, AI could certainly be used to score blog posts, Twitter and Facebook posts, and even the content marketing blogosphere could use a little ML love.
This could also force content marketers to up their game. Writers trained on AI-generated content will have a hard time writing their own copy, as they’ll struggle to write a story that’s as compelling, if not better, than what the AI wrote.
But perhaps the most interesting implication of AI-generated content is the impact it could have on the wider world of content marketing.
My colleague and Gartner’s in-house content marketing expert, Nicole Greene, says that for marketers, the real opportunity lies in the way AI will help marketers create snippets of content associated with journey mapping. “AI is currently being used to create content fragments that can be assembled and tested against segments and personalized experiences. These improvements in creation and optimization suggest that the use of generative AI to create content that supports journey mapping is not far in the future.”
How about use cases outside of marketing? Tim Dooley is part of the data science team at Gartner. He tells me his team crafted a GPT-based tool (the same deep learning model that supports Sudowrite and Jasper.ai) to assist with text generation, text classification and clustering, zero shot learning, information retrieval and chatbots.
According to Tim, “GPT models, and transformer models more generally, allow us to understand language systematically in ways never before possible. They have sped up our internal classification and generation processes immensely and have allowed us to provide remarkably accurate results quickly to stakeholders.”
What’s Next For Generative AI?
I’ve written about AI in years past,1,2 and while back then I felt comfortable knowing we were still a ways off until AI would supplant writers and other content creators, these days, I’m not so sure. But perhaps this is just the natural order of technological progress.
And it’s not just the written word that AI is disrupting. Other generative AI projects like DALL-E 2 and Google’s Imagen rely on the same OpenAI model to revolutionize the way digital images are created. In a conversation I had with Gartner colleague and analyst Kyle Rees about this very predicament, he articulated, “projects like DALL-E 2 acknowledge a truth that authors, artists, musicians — creators — have been struggling with for years: in an era of industrialized information and data, thoughts and ideas are commodity products.”
This is why the future of content creation is so important, and why AI-produced text and video will not replace professional content creators, but rather enhance their capabilities. AI helps with the grunt work, freeing up time for writers to focus on crafting compelling stories, or for video editors to play with cutting-edge tools to tell compelling stories.
The reality is that creativity is an inexact art, and the best creatives are the ones who can string together many different skills to create something new and interesting. AI aids in this process and makes it easier.
To round out this blog post, I’m going to recruit the help of AI to whip up an image. For this task, I’ll use DALL-E mini, an open-source text-to-image generator that was inspired by the DALL-E 2 project I referenced earlier. You’ll see that if I plug in “AI typing at a computer” as a prompt, the AI tool generates a few options to choose from.
While none of these are individually good enough for me to use in a high-production or client-facing deliverable, they’re surprisingly not that bad given that they were generated in about 60 seconds. The real utility in this output, in my opinion, is the way the tool was able to help me visualize and brainstorm an ambiguous concept. If I wanted to, I could pass along one or two of these AI-generated samples to a graphic designer as an idea of a graphic I want custom-made. Thus, generative AI can be used to accelerate time-to-value when it comes to generating new content.
While this example might not be as useful in final production, if I were to be a bit more specific with my query, I get something more usable. Check out what happens when I plug in, “doughnut with sprinkles on a blue plate” into DALL-E mini’s interface. It’s not a stretch of the imagination to think that someone could use one (or all) of these AI-generated images in something like an ad campaign, company website or social post.
Thanks for reading my blog post. I hope you found value in it.
I’m curious…What do you think the future of generative AI holds for marketers, other business professionals and creatives? Share your opinions in the comments section below. I’d love to hear your thoughts.
This post originally appeared on Gartner for Marketers Blog