written by
Dayana Mayfield

We Put Writing AI to the Test for a Blog Post

Content Marketing Marketing Automation 18 min read

In recent years, we’ve seen a huge upsurge in the use of artificial intelligence (AI) and machine learning. However, one area where AI has had little impact is writing content for blogs. In fact, many people don’t know that AI can be used to write blog posts at all.

🤖If you guessed that the above introduction was generated with writing AI, you’d be right.

(Now back to the human.) Writing is a ton of work. It takes an average of three hours and sixteen minutes to write 1000 words.

It’s no wonder that many marketers and business owners are wondering if AI can live up to its promises.

In this post, we’re putting artificial intelligence to the test.

Table of contents:

CTA image

What is writing AI?

Writing AI is artificial intelligence technology that analyzes large data sets in a certain language in order to produce written content in desired formats.

Most writing assistants today offer short-form copy like headlines, bullet points, and short paragraphs. They require some direction. The input is usually sentences and phrases meant to guide the AI’s output towards something relevant and useful.

Writing AI is used for several different functions: marketing copywriting, novel writing, essay writing, and even editing.

How does AI writing actually work?

Most AI writing assistants use GPT-3 to produce human-like text. The AI guesses what to write next based on all of the text it has crawled in the past. Think of it like supercharged predictive text.

The downside of this, especially for bloggers, is that the AI is guessing what words should come next based on what blog content is already out there—meaning that the AI will produce content that is too similar to what is already out there, or will piece together words that go well but that might not yield factual statements.

Can writing AI help with blog post writing?

As content marketers, we know that blog writing is one of the most tedious, time-consuming activities.

While help with email headlines is nice, real help with blog writing would be an absolute godsend.

Why AI can’t help with AI blog writing

Here are the top reasons why AI can’t help you write blogs:

  • Google doesn’t love artificial intelligence blog writing, because it is based on existing content, meaning you’re just publishing the same information with nothing unique.
  • AI writing assistants don’t do a good job at writing within a brand’s style and voice guidelines.
  • AI writing assistants can produce incorrect information, and lazy writers might be tempted to just publish it without fact-checking for accuracy.
  • AI writing requires a ton of editing. For talented writers, this might actually take more time than just writing from scratch.

Where AI can help with AI written blogs

And here are some ways that AI writing can be helpful:

  • Headline and subheadline ideation.
  • Outline ideation, and important topics to cover (this can be with a writing AI or with an SEO research AI like Frase, MarketMuse, or Clearscope).
  • Rephrases content sections that you want to re-use, for example if you’re launching a similar set of articles like competitor alternatives round-ups or templates, you could use the writing AI to rewrite some paragraphs and then edit to make sure it’s unique enough.
  • Spinning up paragraphs and sections when you get writer's block. If you use this strategy, make sure that you enter unique information (that you know from experience or that you researched) into the AI writing prompt so that the AI writer generates unique content that Google will love, instead of just rehashing things from the web.
CTA image

Using an AI blog writer in light of Google’s Helpful Content Update

We’ve all seen copycat content. One blog post covers a topic with “What is...?” and “Why is it important?” and “Examples of...” and then a dozen other blog posts all copy the exact same format.

In some instances, Google does a bad job of showing similar results on the SERP page. In other instances, Google offers a variety of different types of content on the same SERP page.

Google wants to get better at doing the latter.

Just a month ago, Google Search released one of their biggest updates yet, the Helpful Content Update.

This update intends to:

  • De-prioritize content that seems to be written for search rather the humans
  • Unrank content that copies or “skyscrapes” existing articles
  • Unrank and penalize AI-generated content
  • Rank more unique and helpful content

It sent ripples through the SEO community, with many bloggers and content managers worried about the state of their rankings, while plenty others were unconcerned—feeling that they’ve been prioritizing unique content for a while.

In this blog post, we gathered even more tips on Google’s Helpful Content Update and SEO best practices for 2023.

So what does this update mean for AI blogging?

This means that you should not use AI blog writers freely or without constraint. You shouldn’t allow them to generate every paragraph in your article. In fact, you should probably only use AI for 20 to 30% of your article to ensure that the majority of it is unique content.

If you do use AI, make sure to enter plenty of details, facts, and unique perspectives in the prompt so that it creates content that isn’t just a copycat.

This also means that you need to work harder at creating unique outlines with H2s and H3s that other articles don’t have.

Putting Jasper.ai to the test

Can writing AI really produce quality content?

Let’s find out.

We used a writing assistant named Jasper.

Here are the different things we tested with Jasper. Keep reading for the results.

  1. Blog post topic ideas
  2. Blog post introduction paragraph
  3. Blog post outline
  4. Blog content improvements
  5. Social media promotional copy
  6. Email newsletter headlines

Test #1. Blog post topic ideas

First, we’re checking out how Jasper can help with blog post headlines.

topic headlines

Here’s an example of writing AI output:

12 Ways to Improve Your Content Collaboration

Successes:

The topic ideas are pretty good actually. The output of five headlines gives us three listicles that require little editing to be useful.

Failures:

The third output (What are some of the main content management issues that your company aims to solve?) seems more like an FAQ for handling sales objections. However, we can easily change this to read more like a blog post, such as How to Solve Common Content Management Issues.

Our overall opinion on the result of the test:

AI can give you some great ideas for blog posts, but you’ll need to change the “product description” to the general topic you want to cover, to generate new ideas each time.

Test #2. Blog post introduction paragraph

Next, let’s see how the AI does when we ask it to write our blog post introduction. We can select the number of outputs to get different options to choose from.

Jasper AI blog introduction test

Here’s an example of writing AI output:

Starting a blog post with an interesting intro paragraph is tough. Writing one manually can be time consuming and tedious. Finding the right topic to write about can also be difficult if you don't have many ideas or don’t know how to generate more content ideas for your business. Fortunately, there are solutions that make writing easier and quicker than ever before - using artificial intelligence (ai) in the form of chatbots!

Successes:

The first three sentences in the above example are simple and to the point. Unlike common fears about AI, these don’t sound like they were written by a robot.

Failures:

While the first three sentences sound human enough, it is odd that all three begin with an -ing word. It creates a slightly unnatural flow.

Also, the paragraph includes “chatbots” which is totally irrelevant to the topic. And worse, there is an unnecessary exclamation point.

Our overall opinion on the result of the test:

When it comes to actual writing, the AI is less helpful than when it comes to ideating. However, with some quick editing, this intro paragraph is usable.

Test #3. Blog post outline

Next up, let’s see how Jasper does with ideating an outline.

Jasper AI blog post outline test

An example of writing AI output for an outline is:

1. What is writing AI?
2. How does it work?
3. Why should I use it for my blog post?
4. Who can benefit from using ai-generated content on their website or blog?

Successes:

Between the five different outputs generated, the AI covers some basics, like the what, who, and why of the topic. The phrases are also pretty much ready to use.

Failures:

Each outline has only three or four points, so it’s disappointing that the outlines don’t get deep. You also can’t add more criteria as input, such as the familiarity level.

Our overall opinion on the result of the test:

All things considered, the outline feature can be used when writing a high-level overview or introduction to a topic. It likely can’t produce an outline for advanced content and thought leadership.

If you’re really stuck in the outlining phase, it does offer some quick wins to get your brain working again.

Test #4. Blog content improvements

While Jasper can’t yet write long-form content, it does offer content improvements. You can add up to 400 characters at a time, and have the text rewritten for you.

Jasper AI content improvement test

Here’s an example output:

Artificial intelligence has made its way into many industries, including the writing industry. Writing AI is artificial intelligence technology that analyzes large data sets in order to produce written content in desired formats. They require some direction; input is usually sentences and phrases meant to guide the AI’s output towards something relevant.
Most writing assistants today offer short form copy like headlines, bullet points, and short paragraphs.

Successes:

The writing AI has produced clear and informative copy. Unlike some of the other features where the outputs are very similar, the content optimizer produces very different options to choose from.

Failures:

Some sentences are a bit awkward or strangely worded, and the first option reads more like an introductory paragraph rather than an improvement on the input.

Our overall opinion on the result of the test:

If you have a blog post that you want to repurpose into a different post on a similar topic, this might help speed up some of the rewritings, so you’re not posting duplicate content.

Test #5. Social media promotional copy

So maybe AI can’t help you write an entire blog post. What about the promotional copy required to share the post on social media? Let’s see what happens when we use Jasper’s social media feature.

Jasper social media promo copy test

Here’s an example output (the only one somewhat usable):

Writing AI is very much in its infancy. Are you using a tool that will stand up to this brave new world? Or are you writing it yourself and hoping for the best?

Successes:

Unfortunately, this is the most disappointing use case so far. I had wrongly assumed it would be the best at writing social media copy, but alas, that’s not the case. The only good thing here is the ideas generated, not the copy. We can take this and use it to inspire posts with a definition of writing AI, what skills writing AI offers, and the cost-effectiveness of writing AI.

Failures:

The first post isn’t related to writing, the second is incomplete, the third is falsely confident, the fourth is so-so, and the fifth is completely crazy.

Our overall opinion on the result of the test:

Writing AI can’t generate social media copy for you (sorry). But it can offer some ideas. However, the cost of the subscription might not be worth it, since you can easily find prompts and ideas online.

Test #6. Email newsletter headlines

And lastly, let’s see how Jasper might be able to help produce some email newsletter headlines that could be used when sharing a blog post.

Jasper AI email newsletter headlines

Here’s an example of what Jasper generated:

Why you should hire a human writer

Successes:

For email headlines, the AI has generated some that could be usable (Artificial intelligence for writers, Artificial intelligence is writing the content you need). It’s also pretty hilarious that it told us to hire a human writer. 😉

Failures:

The subject lines are all pretty basic and boring. They don’t offer a unique spin on the subject, and they wouldn’t position a company as a subject matter expert.

Our overall opinion on the result of the test:

The subject lines it generates are so obvious that they likely aren’t worth paying a subscription for.

CTA image

Putting Copy.ai to the test

Now, we’re going to test a different AI writer in order to gauge its effectiveness and see whether or not its outputs are any better.

Let’s test the AI-generated outputs of Copy.ai, another popular tool.

We’ll still write about AI writing, partly to be super meta and partly to better compare the tool with Jasper.

Test #1. Blog title

For our first test, we’re generating some blog titles. I input this as the description of the blog topic in order to generate the titles, “AI writing, all about AI writing, everything you need to know about AI writing, how to write with AI.”

Copy.ai blog title test

Here are some of my favorite headlines:

The AI Writing Companion: How To Write With AI
5 Things You Need To Know About AI Writing
How To Write With AI For 80% Of Your Writing Tasks
How To Get Started With AI-Based Writing Tools

Successes:

I was pretty surprised at how great some of these titles are. I like how it threw in that 80% stat. While not accurate, it does make for a great headline.

Failures:

There are plenty of silly titles here, like “Who I Am and What's the AI Writer,” which just proves that AI writing is unpredictable.

Our overall opinion on the result of the test:

Titles are probably what AI writing does best. It doesn’t matter that some of the titles are garbage. All you have to do is pick the ones you like best. But when writing paragraphs, the garbage text can take longer to edit than writing from scratch.

Test #2. Blog outline

For the blog outline, I used a similar input as above, except that I added steps in as well: “AI writing, all about AI writing, everything you need to know about AI writing, how to write with AI, steps to writing with AI.”

Copy.ai blog outline test

Here’s an example of the output:

Section: AI, or artificial intelligence, is a computer program that can solve problems or complete challenges.
Section: We are in a transition period, the transition from human writers to AI-powered writing.
Section: To get started with AI writing, you need to know your topic and tone.
Section: You can use artificial intelligence to write marketing copy about products and services for your business.
Section: You can use artificial intelligence to write blogs for your website.
Section: Come up with a topic.
Section: Brainstorm ideas for sections.
Section: Go to ContentStudio's AI Writing feature and enter the topic, tone, and sections.
Section: Click the Write button at the top of the page.

Successes:

There’s not much to celebrate here. Copy.ai generated 4 different outlines for my query, but none were usable.

Failures:

I was surprised by how poorly Copy.ai performed for the outline test. I would’ve thought that the output would’ve more closely resembled an outline, but most sections look more like sentences rather than headlines.

Our overall opinion on the result of the test:

You may get some ideas from the outline generator, but this AI assistant can’t actually write your outline for you.

Test #3. Blog intro

For this test, I used the same input as Test #2.

Copy.ai blog introduction paragraph test

Here’s an example of the output:

What is AI writing? If you don't know, you've come to the right place. We'll cover different aspects of AI writing in this article. From how to write with it, to what AI writing actually is. Whether you're interested in using AI as a writing tool or just want the basic details behind it, this article will give you everything you need to know about AI writing.

Successes:

The above output certainly isn’t terrible (well, depending on who you ask). It does a great job with simplicity, ease of writing, and positive sentiment.

Failures:

The intro paragraph is pretty cheesy. It also doesn’t include any facts or statistics, which is commonplace in B2B blogs to establish authority and set the stage for a high-quality article.

Our overall opinion on the result of the test:

If you’re writing B2C content or your brand doesn’t really have an established brand voice and you prioritize productivity over style, you could use AI for writing a blog intro. Just be sure to not overdo it for paragraph text, otherwise, the whole article will read like copycat content.

Test #4. Blog paragraph

For this test, I used Copy.ai’s Freestyle feature, which can be used to generate longer paragraphs.

For the prompt input, I wrote, “Why AI writing is helpful and what it can be used for.” I wanted to see if it could write a paragraph about the benefits and use cases of AI writing.

Copy.ai blog paragraph test

Here’s an example of the output:

AI writing is a tool that can be used to automate the process of writing. It uses algorithms to generate content according to a set of parameters, which allows it to be used for many different purposes.
AI writing can help you improve your website's SEO, it can allow you to predict customer behavior, and it can even help you write your next novel!

Successes:

The writing sounds mostly natural, although it’s clearly not super high-quality writing.

Failures:

The second paragraph isn’t factual. While AI can help you predict customer behavior, writing AI cannot.

Our overall opinion on the result of the test:

The results of the test were better than I expected, but a lot of the content came off as really salesy and cheesy. I would recommend not using AI to write paragraphs, unless you’re stuck with writer’s block and would rathoutput iser edit a few paragraphs than write them from scratch.

Test #5. Sentence rewriter

For this test, I asked the AI to rewrite its own sentence, “AI writing is a tool that can be used to automate the process of writing.”

Since that’s a short sentence, I was curious to see if the AI would produce anything more detailed or longer.

Copy.ai sentence rewriter test

Here’s an example of the output:

AI writing is a tool that can be used to create writings that are indistinguishable from those created by humans.

Successes:

There are many options that match the sentence input in style and length.

Failures:

The outputs are pretty bland, but again if you care more about saving time than writing with style, you may not mind.

Our overall opinion on the result of the test:

Rewriting is one of the better uses of AI. But be careful when rewriting someone else’s content as it likely won’t pass plagiarism checkers. If you do use AI for rewriting, you should only do so for 30% or less of your blog. The majority of your blog should still be unique.

Top writing AI apps

Grammarly home page

There are many AI-powered writing assistants on the market today. Here are some of the more popular companies that use AI to help customers write better:

  • Grammarly - Grammar and spell checker
  • StoryChief - SEO optimizer and readability optimizer with specific suggestions
  • ProWritingAid - Similar to Grammarly, but designed for academic papers
  • Skillroads - AI-powered resume writing and LinkedIn profile editing
  • Textio -Augmented job listing writing
  • AI writer - Gathers resources and information for writing articles
  • Articoolo -Helps with researching and rewriting articles
  • Essaybot - Social media writing tool
  • Frase - Analyzes search results to help when writing SEO content
  • TapRecruit - Revises job summaries and listings
  • WordAI - Writes article paragraphs

So, what can writing AI really do in 2023?

Now the question is, what can writing AI really offer today?

  • It can offer ideas to help you write better content.
  • It can rewrite sentences and paragraphs for you to re-use.
  • Plus, it can give you a starting point. You can edit and add to what an AI writing assi,stant produces. (However, that may take you longer than just writing the blog post.)

What AI absolutely can’t do is write a high-quality, well-researched blog post that positions your company as a thought leader.

What is the future of writing AI?

In the near future, we’ll see even more companies starting to use AI for content improvements.

Many companies are still not using AI for editing and revising, which is what AI is currently the strongest at.

For example, a readability and SEO optimizer can show a writer where to make improvements in their article.

SEO AI checker

And in the long run, AI will improve when it comes to writing content. For simple, introductory blog posts, AI might become usable in just a few short years.

However, for very high-quality thought leadership, don’t expect AI to be of help anytime soon.

StoryChief is a content marketing software that brings your collaborators and content channels in one place. Learn more.

Join over 10.000 marketeers for the latest content marketing insights.
Sign up for our newsletter