Human nature is what it is; people use technological advancements for good and bad purposes. In the case of natural language generation (NLG), recent developments enable it to mimic the way humans write.
On the one hand, this technology could create an infinite amount of vacuous content with little purpose other than to generate traffic and advertising revenue. On the other hand, NLG can leverage the human experience and help create rich content experiences we may otherwise never create.
In this article, we explore both approaches to natural language generation. Let’s get started!
AI and The Content Apocalypse
One theory holds that NLG systems have the potential to seriously devalue content and destroy the search experience. Here’s why.
Nefarious operators could use advanced NLG methods to inexpensively generate limitless numbers of pages filled with content spam. The idea is that NLG is used to create content with little human input, other than a keyword phrase to guide its creation. These articles would be stuffed with relevant keywords but devoid of anything but superficial meaning.
The big fear is that these pages will clog search results due to Google’s inability to discern the difference. That concern is justified and understandable. But if this sounds like deja vu all over again, you’re right!
We went through it a decade ago with the commercialization of article spinners. For those unfamiliar with the term, an article spinner takes an article and creates a ‘new’ version by replacing words and phrases, Mad-Lib-style.
Admittedly, these ‘spun’ articles were horrible. But they were good enough to fool search engines. At least for a short period before Google released an update explicitly targeting these low-quality sites. Since that time, Google has continued to address the issue of content quality.
Some fear the same situation could arise with more advanced NLG technology. These people wonder whether algorithms can be developed to deal with this situation.
At present, I think there is little about which to worry. The idea got a lot of traction with the press because they always love a good story. But that’s all it is. Based on my research, AI-generated content won’t break the web anytime soon.
I’ve also experimented with some of the leading natural language generation models. NLG can be very useful in specific contexts. But, unfortunately, it’s overhyped.
There is a wealth of resources focused on natural language generation at Google Research. No doubt, they are working at ways to combat this potential scenario.
NLG and The Transformation of Content Marketing
Not everyone believes we’re heading for a content doomsday. Some of us think we can use advancements in natural language generation for better purposes. In this role, NLG isn’t employed to replace writers. Human input is still required to generate content that takes into account subtleties like brand vocabulary and style.
This distinction is important.
NLG is a tool just like a state-of-the-art text editor. Except we no longer use machines merely to edit text created by humans. Now we use them to generate the text to incorporate into our final piece.
The key to making this work lies in the approach. Natural language generation works best for long-form content when humans are an integral part of the process. Used to augment the writing process, NLG can be wonderful.
Through their life experience, humans can provide unique insight and examples, adding color to potentially sterile content. Here’s part of the NLG output generated by MarketMuse First draft on the subject of becoming a substance abuse social worker. It’s driven by a MarketMuse Content Brief that would otherwise be given to a writer.
The result is not yet ready to be published. You’ll want to edit for correctness, clarity, engagement, and delivery. The author may want to animate the article by using references to their own experience or getting quotes from professionals in the field. In some cases, appropriate images may need to be sourced. And that’s just a start.
There’s a lot of work that goes into creating a piece of publication-ready content. Natural language generation is changing the role of the content creator to that of orchestrator. NLG is removing the burden of lower-level tasks, enables writers to focus on those that add higher value.
The premise of the content apocalypse is that there are bad actors who abdicate their role in content creation to that of machines. This is no different than what we saw a decade ago with the use of article spinners. Text generation is more sophisticated using NLG, but the song remains the same.
If I’m right and history repeats itself, there’s nothing to worry about. If I’m wrong and this time it’s different, there are no worries either.
In this situation, trust and authority will become even more critical. Building that faith and power doesn’t happen overnight, so now is the time to double down on content. Start now, and you’ll be in a good position should the content apocalypse ever rear its ugly head.
Written by Stephen Jeske