The Best AI Writing Software in 2022
I’m super excited about AI-powered writing software. You might think that as a writer, I’d be afraid of AI taking my job, but in truth, I think it’s only going to empower more people to create better content over time. That said, AI writing software has ways to go before we’ll be using it to write for clients at Draft.dev.
In this post, I’ll compare four of the most popular AI writing assistants. I’ll be assessing how well they work on the different types of technical content we write, how easy they are to use, and the value for your money.
In order to test each, I used standard prompts and titles for the articles I wanted the AI to write. You can find these at the end of this blog post. But first, let’s talk about what AI writing software looks like today.
What is AI Writing Software?
Computer-assisted writing tools have been around for a while now. I have used Grammarly and Hemingway Editor for years, but unlike the AI writing software that I’ve reviewed here, these tools merely correct existing content.
The four AI writing tools below generate original content using a sophisticated neural network provided by OpenAI’s GPT-3 model. This means that given a prompt and some guidance, these tools can actually write for you…at least in theory.
As you’ll see, the truth is that these tools are far from perfect. While the best ones might be as good as a very low-cost SEO copywriter, they struggle with esoteric or complex topics. They also tend to introduce factual inaccuracies and are often quite repetitive when left unchecked, so you have to keep a close eye on them.
That said, I see enormous potential for these AI writing tools to become assistants for human writers. As you can see in the video I recorded for this post, you can get a lot of value out of these tools if you know how to feed them good prompts and combine them with human knowledge.
Comparing Today’s Top AI Writing Software
In order to better understand the potential and limitations of AI writing software, I decided to test-drive four of the most popular tools out there. I’ll continue to try out new tools as they hit the market, so if you’ve got one for me to try, shoot me an email.
Copy.ai
Copy.ai was the first tool I tried. Copy.ai bills itself as a way for marketers to overcome writer’s block, supporting everything from sales copy to blog content.
In a couple of minutes, I had figured out how to work with it. They have dozens of features in the left sidebar, but it’s not clear what all of them are or how I should use them, so I mostly stuck to the blog titles, intros, and bullet points features.
Writing a roundup worked pretty well. It was able to come up with generic descriptions for each item in my given list, but it struggled to get much deeper than that. I’ve come to find that this lack of depth is typical among AI writing tools, especially in highly specialized topics like technical writing.
Copy.ai was worthless at writing tutorials with code samples, but I could see many of the topic ideation and keyword generation features being somewhat useful if you struggle to come up with topics. My only concern is that the AI isn’t really thinking strategically, but rather coming up with ideas based on existing content. Still, it might help get you “unstuck” at times.
Pricing for Copy.ai is in the middle of the road. After your 7-day free trial, you’ll pay $35/mo for an individual account. Team pricing requires a custom quote.
Jarvis (now Jasper)
Jarvis (recently renamed “Jasper”) is definitely aiming for the premium end of the market, and it’s clear that they’ve spent more effort on the product (and marketing) than many of the others here. That said, the writing itself is not that much different from cheaper options.
On the plus side, it was really easy and self-explanatory to get started with creating a blog post. Jarvis offers several kinds of content, but as I was primarily focused on long-form blog posts, that’s where I spent most of my time. Unlike Copy.ai, you can’t pick the text that goes into your post though, so I found that I was doing a lot of deleting when an AI-written passage made no sense. Jarvis also seems to randomly add extra line breaks to content (or maybe this is a flaw in GPT-3?), which it seems like it could strip out for me to save time.
As for writing technical content, it was fairly good at writing introductory content for the roundup. From what I could tell, the descriptions of each tool I was comparing were accurate, but it struggled to get into much depth when I prompted it to create a technical guide and it couldn’t make progress on the tutorial I tried to have it write.
As I mentioned, Jarvis is at the premium end of the price range for AI writing tools at $60 per month, and unlike most of the other tools I tried, there’s no free trial. While it might perform better on other kinds of writing, I would not recommend it for blog posts at this point as there are cheaper and easier-to-use options out there.
Frase
Unlike the other items on this list, Frase is more explicitly focused on helping you create SEO-driven content. It analyzes the existing search engine results for the keywords you give it to help you create a brief and outline.
Like the other software here, Frase does include an AI writing assistant as well. It seems that Frase is trying to integrate both the research and writing work for SEO writers in one place, and I think it does a fairly good job of it.
That said, it wasn’t any better at creating technically detailed or especially interesting content. Like the others on this list, it tends to work best on high-level overviews of non-technical topics or low-depth listicles. That said, if SEO writing is your primary focus and you don’t need an AI writing tool to help with other forms of copywriting, Frase might be a good option.
You can get a 7-day trial of Frase for $1 and then pay $19 per month for an individual plan after that.
Copysmith
Copysmith embraces an AI-assisted workflow, expecting you as a writer to fill in the gaps and create most of the content yourself, but layering an AI that will suggest copy for you as needed. It’s most similar to Copy.ai above, but initially, I found Copysmith a bit harder to use. It was a little less clear if you could write long-form content and the blog post idea feature initially gave me very strange results.
That said, I’ve found that Copysmith really grew on me the more I stuck with it. I’m now using it on a daily basis to generate article titles and descriptions based on keywords, and my team is starting to integrate it into our content outlining workflow. The most defining feature is that it generates multiple options for each request you submit, so instead of clicking “regenerate” over and over, you can pick from a long list of automatically generated pieces of text.
Copysmith is reasonably priced at $19 per month for individual accounts, and they’re starting to integrate with other tools like Zapier. I’m really excited about how much this could improve our content planning and writing workflows at Draft.dev this year.
Conclusion
The options for AI writing software are growing and improving. While many use the same underlying model (GPT-3), we’re likely to see competing models emerge and better implementations built as the technology matures.
We certainly won’t be replacing our writers with AI anytime soon though. The truth is that AI writing software is a versatile assistant, but far from replacing high-quality technical writers for the kind of content we produce at Draft.dev.
That said, I could see AI writers starting to replace some low-end freelance writers or at least make writers more efficient. Like many other industries have seen, artificial intelligence is a great supplement to human intelligence.
Appendix
Prompts Used - Here are the three prompts I used in case you want to test these tools on your own:
Best continuous integration tools (roundup)
In this roundup, we’d like to compare 10 or more of the most popular continuous integration tools available. In the end, we’d like to help software developers pick the best tool based on their budget, stack, and preferred workflow.
Introduction to Redis (guide)
In this blog post, we want to offer software engineers a high-level introduction to Redis, the key-value store. We also want to provide code samples to show developers how to use it and what commands are available.
How to Create Your First NodeJS Application (tutorial)
In this tutorial, we’d like to show new software developers how to get started with NodeJS. They will build a small application with a single page that says “Hello, World!”
By Karl Hughes
Karl is a former startup CTO and the founder of Draft.dev. He writes about technical blogging and content management.
Want ideas for your next blog post?
Kickstart your content marketing efforts. Download our list of 50+ winning ideas for your company's blog today.