isotropic-2022
Share
Blog
Search
Menu

Tips And Tricks On How To Humanize AI Content In 2023

By James LePage
 on August 9, 2023

Tips And Tricks On How To Humanize AI Content In 2023

By James LePage
 on August 9, 2023

AI content is relatively detectable. In this article, we'll cover how this content is generated, the telltale signs that it's written by a LLM (large language model), and how you can humanize AI content.

Before we cover how are you can humanize AI writing, it's important to understand how it's actually created. By knowing about the technical mechanisms behind the words generated, you'll be able to better understand how to combat the markers that it is written by computer and not human.

How AI Generates Content

AI content is made using big datasets, complex algorithms, and deep learning. One of the most powerful kinds of AI used today for generating content is called large language models (LLMs).

chat_gpt_writing

GPT-4 by OpenAI is an example and these models get trained by feeding them insane amounts of text - a ton of sentences, paragraphs, even full articles.

Over time, the model starts to pick up on patterns in how people write.

how LLMs determine the next word
Source

It's like teaching a little kid to talk, but way more advanced. Once the model is trained up enough it can start generating content based on whatever prompts you give it.

Catching the AI

Figuring out if something was written by AI takes a careful eye and knowing some of the classic AI slip-ups. Here's what to look for:

  • Perfect Grammar: AI often misses those tiny grammar mistakes that humans make. The writing might feel too clean and formal.
  • No Personal Touch: LLMs don't have real experiences or emotions. So the content wont have any personal stories or authentic human feeling.
  • Redundant Rambling: Sometimes AI just uses way too many words to explain simple ideas. There's a lot of repetition. AI content sometimes just seeks to fill space, without adding legitimate value.
  • Limited Knowledge: If the content only talks about events or facts up until a certain year (like 2021 for GPT LLMs from OpenAI), that's a sign it could be AI-generated.
chat gpt training cutoff
The 2021 knowledge cutoff

There are also several AI detector tools on that market, which can accurately detect writing generated with ChatGPT, etc. Originality AI is one of them. For others, read this article:

How To Humanize AI Content

Bridging the gap between robotic machine text and real human writing is what humanizing AI content is all about. It brings a 'realness' to articles, blogs, etc. As AI becomes more common for pumping out content, keeping that human touch is key for reader trust and interest.

Change Up How You Prompt the AI

Having an AI make content usually starts with the prompts you give it.

A direct prompt can get you a stiff response. Instead of "write a blog about climate change" try "key points on people's effects on climate change" and you'll get a more nuanced set of ideas. From there, shape those points into a story that connects with human readers - a collaborative piece combining AI speed and human touch.

Using AI for brainstorming instead of just making content lets things flow more naturally in the end result. It becomes a tool for inspiration, not just production.

Learn more about prompt engineering here:

Vary Your Sentence Structure

One clear sign of AI content is the repetitive sentence structure. Even if grammatically flawless, it often lacks the rhythm and flow of human writing. To fix this manually edit the content, mixing in short punchy sentences among longer descriptive ones.

Creating a mix of different sentence lengths and structures not only makes the text more interesting but also mirrors how people naturally speak and think. This variation can turn a boring piece into an engaging read.

Replace Big and Unnatural Words

AI tries to be precise but sometimes uses fancy vocabulary that isn't relatable. Words that arent everyday language can distance readers. Instead, consider substituting these words with simpler, more common synonyms.

For example if the AI uses the word 'utilize', it might be more relatable to just use 'use' in many cases. Making these small tweaks ensures the content connects with more people and feels more conversational.

Break Standard English Rules

While it's important to be clear, adding in slang, idioms or regional phrases can make it sound more real. AI tends to follow proper grammar rules, but humans often play with words, creating a unique voice.

Introducing a tiny bit of informality—whether it's a popular saying, cultural reference or playful spin on a traditional phrase—can make readers relate to your content more, making it feel like a chat with friends rather than a soulless lecture.

Include Sources from Recent Years

LLMs like GPT-4 have a cutoff point for what they know, meaning they might miss things that have happened after that date. To keep your writing up-to-date, you'll need to add in new sources facts, or events yourself.

Mixing the AI's huge knowledge base with what's happening now, not only makes your writing more believable and accurate, but also ensures it relates to today's readers.

Nothing humanizes writing more than personal stories or experiences. Those moments - memories, lessons observations - add a depth no AI can recreate. Even if an LLM writes most of your content, sprinkle in those personal touches.

Check The Output!

We've all seen the memes of students submitting essays with the heading of "As a LLM by OpenAI, I can't answer...".

ChatGPT student essay meme

ChatGPT is a chatbot. That means that it will sometimes interspersed responses of its own at the beginning or within the content. If you don't proofread the content and simply copy and paste it, it's very likely that there will be telltale signs that it's ChatGPT that is conversationally responding.

Don't be lazy!

Use AI Humanizer Tools

Funnily enough, there are several tools that leverage AI to convert AI content into more human writing. They have their own internal models, processes, and systems, to take the raw output from ChatGPT, and rewrite it to sound much more human allowing it to pass AI detectors and the 'eyeball test'.

If you're interested, we wrote an entire article on this niche of "undetectable AI tools":

Conclusion

The point of humanizing AI isn't to distance yourself from technology, but to blend AI efficiency with human emotion. It's that fusion that creates content that informs and connects. As we move into an increasingly digital world, keeping that balance will be key for writing that impacts.

While AI, especially LLMs, have revolutionized content generation, it's essential to understand the difference between machine-generated and human touch content.

As the line blurs, being aware and knowing how to humanize content ensures that we maintain authenticity and connection in the digital age.

Whether you're using AI for efficiency or out of necessity, integrating a human touch can make all the difference in the reception of your content.

Do you have any tools or suggestions of your own? Feel free to leave them in the comments section below, we'd love to hear your thoughts.

Subscribe & Share
If you liked this content, subscribe for our monthly roundup of WordPress news, website inspiration, exclusive deals and interesting articles.
Unsubscribe at any time. We do not spam and will never sell or share your email.
Subscribe
Notify of
guest
0 Comments
Inline Feedbacks
View all comments
Article By
James LePage
Contributors/Editors
notloggedin
James LePage is the founder of Isotropic, a WordPress education company and digital agency. He is also the founder of CodeWP.ai, a venture backed startup bringing AI to WordPress creators.
We're looking for new authors. Explore Isotropic Jobs.
linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram