AI content is HOT.
Everyone’s talking about AI content now, whereas a few years ago, people were just curious about it.
People are doing studies on it, so I want to give EditorNinja’s official take on AI content.
I don’t like claiming superlatives that I’m not confident in, such as that we’ve edited the most AI content of any service on the Internet. But I can honestly claim that we’ve edited 7 million plus words of AI content, so we have a few lessons to share.
AI Edits Take Longer
The first thing we’ve learned about editing AI content is that editing takes 50%+ more time than editing human-generated content.
This depends somewhat on the editor and the quality of prompts. Improving your editing processes and prompts may result in some efficiencies (5-10%), but humans are still needed.
Editing AI content doesn’t take double the time, but it undeniably takes more time when done well and requires a different skill set than traditional editing.
Editing AI Content Blends Editing + Writing
Editing AI often feels like a mix of editing and writing. It’s not just fixing errors; it’s adding depth and personality.
You’re rearranging, fixing tone, adding details, stories, and stats. Editing AI content is similar to a traditional heavy-duty edit, but it’s closer to rewriting and creating a second draft that is then ready for an editor.
There’s No Human to Give Feedback To
This is the crucial point that many people miss about AI content versus human-written content.
When a human writes content, you can send it back to them to make changes and revisions. Traditionally, this was the editorial process. The human would write, the editor would give deep feedback, a second draft would be done, the editor reviews and gives feedback or copy edits while proofreading, and then it’s sent to publishing.
With AI content, the process becomes much simpler but requires much more of the editor. Often, the creator is also a writer and the editor.
Someone creates the brief, then uses AI to generate the content. Ideally, they’re doing it section-by-section, but many creators try to generate the whole article at once. Then they take the output and work with it heavily. They’re rearranging pieces, fixing tone and voice, fact-checking, adding stories, and all the other things that writing involves.
Then they still need to copy edit and proofread to catch any mistakes that they introduced.
The editing phases are now the vast majority of the work.
Personally, as a writer, I think this kills some of my creativity and makes the creation process less enjoyable. However, this process is more scalable and faster than the old way. Net-net is positive, but there are downsides.
Why Do AI Edits Take So Much Time?
AI gives you the basics – facts, structure, etc. But it’s often “mirror content,” which sounds good but lacks real substance, perspective, or a personal touch.
Mirror content is content that, at first review, says something really good, but when you dig deeper, it’s not.
The bulk of the content is there, the facts are there, and most of the structure is there. There are words on a page that make sense, but it doesn’t have a perspective. It doesn’t have a “take.” It’s just regurgitating information.
Improving mirror content is where the bulk of the AI editing work lies.
Tell-tale Signs of AI Content
There are a few tell-tale signs that something is more likely to be AI-generated than human-written. Good editing makes it hard, or impossible, to tell the difference because these have been edited out.
The “AI Feeling”
AI content has a distinct vibe. It lacks a personal touch, depth, and the variance of language that makes good human writing so interesting.
The reader might be unable to put their finger on it, but when asked, they’ll admit that something “feels” off. And they’re right. It’s mirror content.
Human-edited AI can help remove that vibe.
No stories or experience
AI content lacks real stories or lessons from experience. That Expertise in EEAT, otherwise known as subject-matter expertise? That’s where the human touch comes in.
These can be stories directly from the author, from the person who is being interviewed, or simply examples that you find on the Internet and retell (and link to the source, of course).
AI can tell the facts. But it can’t tell stories from experience.
First vs. Third Person
This is one I don’t see discussed often. AI content is usually written in the third person (“statistics tell us”), while human writing is more often written in the first person (“when I did this”). Writing in the first person feels more real.
Thus, to “fix” AI content, especially editorial content, changing third-person to first-person can be a big step towards “humanizing” it.
We Love AI, But It’s Only a Good Start
AI is great for drafting, brainstorming, and getting feedback on writing. I even used it recently to get feedback on the introduction of a book draft I just finished. After about an hour of back-and-forth with Gemini (my tool of choice these days), I emerged with a much stronger, rearranged introduction.
AI is a tool, but it’s not a replacement for human work. I used AI to help me with this article, but then also put in another couple of hours of work to make it say what I want.
AI Isn’t A Panacea
The gurus and Instagrammers will tell you that AI will take your job and you can replace your whole team with AI “agents” (don’t even get me started).
The truth is, you can’t just get AI to do all the work for you. It’s a tool that needs a trained human.
Just ask Klarna. According to this article, Klarna went too far by pausing hiring on its support teams and trying to use AI instead to provide service. After a $40B decline in company value, mainly from the pandemic boom and then a cooling off of consumer spending over the last few years, the CEO said:
“As cost unfortunately seems to have been a too predominant evaluation factor when organizing this, what you end up having is lower quality,” he said at Klarna’s Stockholm headquarters.
“Really investing in the quality of the human support is the way of the future for us.”
While their story isn’t typical, it’s cautionary. AI is a great tool, but humans can take it too far. Klarna swung the pendulum too far.
What We Look For When Editing AI Content
Ok, let’s get tactical now. What specifically do our editors look for when editing AI content, so that it’s “humanized” and ready to publish and drive business results?
Here’s a rundown:
- Fact-check for AI hallucinations and fix them
- Check for repetitive sentence structures and simplify them
- Make the language sound less robotic and more humanlike (third person to first person, for example)
- Smooth out transitions between paragraphs
- Rearrange content as needed to improve the narrative
- Add stories, examples, and statistics as applicable to build trust with the reader
- Check and replace common/repetitive AI keywords (moreover, array, foster, robust, mouthwatering, etc)
- Fact-check everything again, including links
- Copy edit (format, tone of voice, grammar) according to the style guide
- Proofread (final typo review)
- Recommend (via comment) where images may help communicate the ideas better (if applicable)
- Check for double spaces and typos.
AI’s most significant issue is hallucinations, so we check for that specifically.
Then, we remove repetitive language to make the writing sing (in keeping with the style guide’s tone and voice, of course). If needed, we’ll rework the copy to be more first-person than third-person, to add that human element.
We thoroughly read and rearrange content to improve the flow and narrative.
We add in trust elements like stories, statistics, and examples. If necessary, we link back to the original.
We do another fact check.
We copy edit everything according to the provided style guide. AI can be helpful here, but we’ve done it by hand thus far.
We do a final proofread. Some people will tell you that proofreading isn’t necessary with AI. Still, we’ve found that it is needed, especially when you’re properly humanizing and changing tenses from third-person to first-person. A proofread can save you from embarrassing unforced errors introduced by humans (ironically!)
How to Use AI to Edit
A common question we receive is whether “editing can just be done using AI.”
The answer is complicated.
You can absolutely use AI to give you feedback on your writing, and you should. Like the intro to the book example, asking AI for feedback can help you rework things into something stronger. If you’re the human using AI to create content, this is a great way to get content to a strong place much faster than before AI was part of your workflow.
You can also use AI and custom projects/GPTs/Gems (every tool has its own naming convention) to ask AI to give you feedback on what needs to be changed to align the copy with your existing style guide. We’re investigating this more and more at EditorNinja, and we’ve seen significant improvements to our content.
Self-Editing with AI is Great With Low Content Volume
Self-editing with AI works great when your content output is low.
If you’re producing 1-2 articles a month, using AI this way can help you get content to a much stronger place. If you don’t have the budget to work with an external editor, a workflow like this won’t necessarily help you produce MORE, but what you create should be better.
The problem is that editing using AI still requires YOU. This is what people miss about editing – they absolutely CAN do it themselves. They’re usually perfectly capable.
But self-editing AI content has the same issue as self-editing human-written content – you can’t scale your time.
You Can’t Scale Your Time Without Hiring
But when your volume increases, capability won’t get you there.
You’re one person with a finite amount of time and focus. You might be able to work 10% more to produce a bit more, but you can’t work 200% more to deliver 3x the content to drive the traffic needed to hit your traffic and lead goals.
When that’s what you need, working with an editor who leverages AI tools can be a win for everyone.
They get the work, you get your time back, and you also pay for less time than you would have in the past. Or, you can pay for the same time and get 50%+ more work done.
Producing AI Content at Scale?
EditorNinja helps companies improve their content as they grow their output. If you’ve reached the point where your team can’t handle the volume of content that is required to reach your goals, get in touch with us to discuss how we can help you out. We’ve edited 7m+ words of AI content since 2023, and we’re actively bringing on new clients to help them grow.