This Blog Sucks: Our Experiment With AI-Generated Content

I consider myself a futurist — one known to pontificate about technology’s rapid evolution and how it might potentially impact my life one day. I also have some patient friends and co-workers. They’ve all had to bear my obnoxious ramblings at one time or another.

I only mention this because it has made me a technological optimist. For instance, I believe I will one day own a self-driving, carbon-neutral car. I believe I will see humankind walk on Mars. I believe vegan cheese will eventually taste good.

All of these things are likely going to happen sooner than any of us are imagining.

I’m not without my doubts, though. Say, for instance, when a company hounds me relentlessly with ads claiming their patented artificial intelligence (AI) can write compelling blogs for me. As a B2B marketer who understands how important the emotional element is in messaging and branding, I tried my best to ignore their claims. Still, the ads persisted — day in and day out.

Finally, I acquiesced. After all, what kind of futurist would I be if I didn’t at least try it? The company (which shall remain nameless) even offered a free trial.

So I took them up on their offer and we tried it out. Here’s what we found.

Can AI write a good blog?

Let’s start on a positive note. Producing the blog was indeed as effortless as advertised. All we had to do was plug in a topic and choose a tone of voice. And within less than a minute, a blog materialized seemingly out of thin air.

Our topic? “Is AI-Generated B2B Content Any Good?” We went with a “professional” tone. Next, we read what it regurgitated. (Spoiler alert: it sucked.)

It didn’t suck in the way we had anticipated, though. It lacked those goofy “robot-isms” you might see on social media where an AI bot allegedly writes movie scripts or song lyrics. It was surprisingly readable and not too grammatically messy.

But it was also slapped-together fluff: redundant, unoriginal and entirely useless paragraphs, nearing the point of plagiarism. Keywords were repeated ad nauseum. (For instance, “AI generated B2B content” was stuffed into our eight-paragraph blog 10 different times.) It also conveniently included around 500 words in total, which is what a lot of SEO “experts” still believe is needed to somehow unlock top-of-page search engine rankings.

In that way, I found the tool I used was even more insidious than I initially had imagined. If you expect to receive something bullet-ridden with grammatical errors and misspellings but end up with something semi-coherent, suddenly semi-coherence reads like Hemmingway. This will likely trick a lot of marketers into thinking “this is good enough”. But simply put, it isn’t.

SEO- and AI-Generated Content (Another spoiler alert: It doesn’t work)

Google has been cracking down on this activity for a long time. Most recently (August 2022), they rolled out an algorithm called “Helpful Content” that is entirely geared at clearing this kind of junk from its index. According to Google, it’s part of an effort “to ensure people see more original, helpful content written by people, for people, in search results”. (Notice the “by people” bit? That wording is intentional.)

This is Google’s algorithm battling another algorithm. It’s literally a robot war. (How’s that for futuristic fantasy?)

An authentic content strategy matters

Before you’re tempted to go down this road, try to remember the big picture. Even if you can trick search engines into ranking this dribble, do you think it’s going to influence your customers? If you operate in a high-consideration purchase environment, you have to ensure that everything you serve your audience is going to influence that consideration in a positive manner. Content within the digital landscape is your opportunity to make authentic connections with your audience and even potential customers. If a user receives content like the kind these AI bots are serving up, it’s going to reflect laziness about your brand at best, distrust at worst.

Contrast this with Nutella’s latest campaign that leaned into AI in an amazing way. With the assistance of an algorithm, the Ferrero-owned brand produced seven million unique product label designs that populated store shelves. The campaign was a huge success and all seven million jars sold. What makes this case different from our blog experiment is that it wasn’t trying to trick anyone. It merely leveraged technology to extend their production capabilities. There’s neither a physical nor profitable way in which a human could manually design that many labels for spreads that cost only $4 a jar. I would argue this campaign is not replacing human creativity as much as it is accenting it.

That’s not to say we’ll never get there with AI mimicking creativity or emotion convincingly. Again, as a futurist and technological optimist, I definitely think it’s possible an AI bot could assist in the heavy lifting of blog-writing one day. Most of us already lean into robotic assistance one way or another. Every time you run a spell-checker for instance, you’re employing the assistance of an algorithm. But an AI bot driving the creative process? I wager we’re at least 10–15 years away from seeing anything productive coming out of that kind of technology. 

We ran a few additional experiments on other subjects just for the fun of it. Our favorites were “Dog Care Tips” with a “witty” tone and “Why Do They Keep Rebooting Old Movies” with an “angry” tone. You can see all of them, including the first experiment below, raw and uncut.