How to Write a 650-Word Blog Post Using Only Artificial Intelligence
Artificial intelligence wrote an entire blog post for me.
(Seriously. All of it.)
I gave AI a topic, and the machine did the rest.
The result was a 100% factual, 100% original blog post that totaled about 650 words. It was about the top seven books on AI and machine learning. The tool I used selected the books, then wrote excerpts about each one—in completely natural language.
AI tools that generate language are nothing new. But the technology appears to have advanced to an unprecedented level. We might not be far away from machines writing in-depth longform content.
Below, I detail my experiment and unpack the implications of advanced language generation AI on the world of marketing.
How We Achieved Successful Content Generation with AI
To create the post, we used a tool called HyperWrite.
According to the website, “HyperWrite uses the world's most advanced AI to help you write better and faster than ever before.”
Unlike some other natural language generation (NLG) tools we’ve seen, HyperWrite appears to have been “trained on much of the Internet,” so it can actually write factually correct content. Sometimes, NLG tools write sentences that sound right, but don’t have real facts or figures in them. They mimic human words, but not human thoughts.
HyperWrite appears to be a different animal.
It provides definitions and information that appears to be completely correct. (You can see this at work in the video below, where I ask it to write about “What is machine learning?”)
For our AI-written post, we told HyperWrite to create a non-fiction post with the following description:
“What are the top books about artificial intelligence that are available today? This post shares the best, most popular and best selling books about AI, machine learning, and data science.”
From there, we gave the machine a title: “7 Top Books on Artificial Intelligence and Machine Learning.”
Then, we turned it loose.
The way HyperWrite works, you click the “ThinkAhead button” and the machine generates a snippet of language. Human reviewers select which language choices they like best out of three possible options.
In this case, I was the human reviewer. In every instance, HyperWrite generated three snippets of language that made perfect sense and were factually correct. The only real selection I made between options was based on making the text sound less repetitive.
As you continue to click the ThinkAhead button, HyperWrite assembles your post.
For our post on books, HyperWrite selected the books and wrote factually correct descriptions of each one. In a few cases, it started to repeat books before finishing a list of seven unique titles. In these instances, we clicked the ThinkAhead button a few more times until it generated a new book.
Even with these hiccups, HyperWrite constructed a 650-word post in minutes—and my only action was clicking the ThinkAhead button.
Once the machine finished writing the post, we ran the text through Grammarly to clean up any language and check for plagiarism. HyperWrite made no typos, though Grammarly did correct a couple phrases to make them fit our preferred style and tone. The post was 100% original.
At the end of the day, the only part of the post written by a human were the final two sentences:
“What's your favorite book about artificial intelligence? Let us know in the comments.”
Here’s the wildest part of the entire experiment:
Each section of the post is formatted with a numbered book title, then lists out the “Author,” “Why you should read it,” and a “Key takeaway” for each title.
We did not tell the machine to follow this format. It came up with it on its own—presumably after looking at how similar articles online were structured.
We’ve followed the marketing AI space for years now, and this functionality still surprised us. It seemed to indicate that AI-powered content generation was finally coming into its own—and that we may be entering an era of fully (or mostly) machine-written content.
What This Means for Marketing
What does this mean for marketing?
It means things are about to change.
Make no mistake, HyperWrite is just one tool. And it has limitations. We intentionally gave it a highly structured list post to write. Given more open ended prompts, it can begin repeating itself quickly.
But, under the right circumstances, it has the ability to reliably generate hundreds of words of natural-sounding narrative almost automatically.
As an observer of the space, that feels like a new milestone in AI-powered content generation to me.
It is not outside the realm of possibility that brands will start creating machine-generated content at scale. If that's possible (and I think it is) then the brands who get there first could conceivably create hundreds or thousands of articles each day or week, and quickly establish dominance in search across any given market.
Combined with smart human writers and sound editorial strategy, tools like HyperWrite almost certainly signal a more intelligent future for content marketing. Even if we can't auto-generate content at scale, we now have reliable tools to dramatically increase output and reduce costs.
As technology gets smarter, marketers have to get smarter, too. We're going to need to understand what these new tools can do—and what they can't. We're also going to need to understand where our prime human talent fits into this equation.
It's very likely we're about to see a future where human storytellers and intelligent machines work hand-in-hand to produce great content at scale.
What exactly does that look like?
I think it's too early to tell. But our experiment shows one possible path.
However, one thing is for sure:
Whatever it looks like, we're about to enter a brave new world.
About Mike Kaput
Mike Kaput is Chief Content Officer at Marketing AI Institute and a senior consultant at PR 20/20. He writes and speaks about how marketers can understand, adopt, and pilot artificial intelligence to increase revenue and reduce costs. Full bio.