How to Detect AI Writing just by Reading

How to detect AI writing? Was this written by AI? Is it possible to detect AI-generated text? These are some of the famous questions people have been asking over the past couple of years. The interesting part is that the questions originated in 2023!

AI content and AI writing have been on the rise for the past few years. Generative AI is getting better at writing human-like phrases. This makes it difficult to distinguish if content is written by a human or an AI tool.

AI writing tools are becoming more popular with each day. We can understand this by simply following the trends.

I opened Google Trends and typed in “AI Content” and selected data for the past 5 years:

You can see the significant increase in terms of people searching for “AI content” on Google.

With a massive number of writers using AI tools for generating content, it has become crucial for us to detect AI similarity.

But can AI be detected? or how can you detect AI writing? The answer is yes! We can detect AI writing. There are two ways that can help us check if something is written by AI.

We can either do it manually or use AI tools for detecting AI content. While our AI detector has recently achieved 98% accuracy, we still can’t blindly trust a tool!

Why check if content is written by a human or AI?

The content can be in different forms, such as assignments, blogs, articles, etc. If assignments are totally written using AI, it is misconduct. It means the student has done no research at all and simply used an AI tool to generate content.

If the content is in the form of a blog or article, it is more important to check whether it is written by a human or not. This type of content is typically used for marketing purposes, and search engines have their own guidelines when it comes to content.

Google (the most popular search engine) recently published its helpful content guidelines. AI content lately is not that efficient and does not meet those guidelines.

Google's September 2022 Helpful Content Update was the first step when they take down websites with AI spam. Although the core update's focus was to reward websites with genuine content, a massive number of websites with AI content got penalized. It means Google can detect AI content and take action if it isn't helpful for the audience.

Now that we understand AI content and why we should check it. Let’s understand some popular ways through which you can detect AI writing just by reading.

Also Read: How to humanize AI content to make it more readable?

How do you detect AI writing just by reading?

If you are someone who has the ability to read and understand what content is and how it is usually written, you can easily detect AI writing just by reading.

1. Read the content carefully

The first and most basic thing to do to detect AI writing is to read it carefully. When you are reading a particular piece of content, try to understand if you are finding it helpful.

If you are reading 100% AI content, you’ll find that it isn’t helpful. It will be straight, emotionless, dry, and boring.

2. Look for Jargon

If AI content were a recipe, jargon would’ve been its main ingredient. Jargon means difficult or lifeless words. AI content uses algorithms to generate content, which is why it contains a lot of jargon.

Example No. 1:

I asked ChatGPT to write me 20 words for “AI content is easier to read,” and this is what it provided me.

ChatGPT’s Content:

AI-generated content manifests with greater legibility, thus simplifying the consumption of complex information for a broader readership.

You can see ChatGPT used jargon in its short answer.

Example No. 2

I asked Google’s Bard to write me 20 words for “AI content is easier to read,” and this is what it provided me.

Bard’s Content:

The cognitive parsing of textual output generated by artificial intelligence proves demonstrably facile for the average reader.

You can see Gemini (formerly known as Bard) also has jargon in its response. It also gave me another draft, but that is also not quite right!

Example No. 3:

Now let me write some common ways a human writer would write approximately 20 words on “AI content is easier to read.”

1. AI content is easier to read because AI tools structure content better than most of the writers.
2. By giving a few prompts, we can make AI content easier to read, and it will become more engaging and effective.
3. You can add a human touch to AI content to make it easier and more useful for users to read and engage with.
4. AI content is easier to read if we compare it with the content of an inexperienced writer. AI content will be more factual and accurate.

These are some of the ways in which a human can write something like “AI content is easier to read.”

Make sure you notice jargon in the content if you are reading. This will help you understand that the content isn’t coming from a writer but instead from an AI tool.

3. Check for Readability

As we discussed above, AI content usually has jargon. AI tools generate content using language models. Those language models are not efficient as of now to make content helpful for readers.

There are over 7,000+ spoken languages worldwide, and English is one of them! The English language connects the world because it is the global language. It isn’t possible that people from Africa would understand English the way people in New York do or people from Indonesia.

Since AI tools can’t understand this and are not that efficient as of now, their content is difficult to read.

If you are someone who has a good understanding of the content and how it is written,. You can read an article and see if you find the content difficult to read along with the jargon.

This will help you understand if the content is written by a human writer or AI tool.

4. AI content has the same structure

This is another common thing that you can notice to detect AI writing. As the AI content is generated by algorithms, it will always have a similar structure. For example, if you are reading an article on technology or AI, some starter sentences will look like:

1. In this digital age, technology has become indispensable.
2. In the modern era, technology is booming.
3. In the ever-evolving era of the digital age, technology is escalating at an unprecedented rate.
4. In the digital landscape of the modern era, technology is climbing to new heights.

If you are someone who understands content, this is a very basic and outdated way of starting a sentence. Look for these sorts of sentences while reading to detect AI writing without using any tools.

5. Excessive Sentences in Passive Voice

The most common way to avoid plagiarism is to turn that content into a passive voice. When you turn content from active voice to passive voice, most tools show you that your content is not plagiarized.

AI tools and their algorithms use this to generate content. AI tools are basically trained on large datasets, so they use existing information to generate content.

Example No. 1: 

I gave ChatGPT a prompt to write me 2 paragraphs on “Why ChatGPT content has sentences in passive voice.”

ChatGPT content:

Content generated by ChatGPT often includes sentences in the passive voice to achieve a certain level of objectivity and formality. The passive construction is chosen so that the focus can be shifted from the doer of the action to the action itself.

This is particularly useful when the identity of the actor is unknown or irrelevant to the context of the statement. By utilizing the passive voice, emphasis is placed on the action or the recipient of the action rather than on the subject, which may not always be central to the conveyed message.

Furthermore, the passive voice is employed to vary the sentence structure, thus enhancing the readability and maintaining the interest of the reader. Academic and technical writing, which often seeks to minimize the appearance of bias or subjectivity, regularly takes advantage of the passive voice.

In such cases, the passive voice is seen as a stylistic choice to communicate information in a way that seems impartial and focused on the information itself, rather than the agent performing the action.

Passive Voice Sentences:

1. The passive construction is chosen so that the focus can be shifted from the doer of the action to the action itself.
2. By utilizing the passive voice, emphasis is placed on the action or the recipient of the action rather than on the subject.
3. The passive voice is employed to vary the sentence structure, thus enhancing readability and maintaining the interest of the reader.
4. Academic and technical writing, which often seeks to minimize the appearance of bias or subjectivity, regularly takes advantage of the passive voice.
5. The passive voice is seen as a stylistic choice to communicate information in a way that seems impartial and focused on the information itself, rather than the agent performing the action.

You can see that the two paragraphs had 7 sentences, and out of those 5 sentences, three were in passive voice.

When you are reading content that has more than 70% sentences in passive voice, it is definitely written by AI.

Also read: Is AI content good for SEO?

6. AI Content Usually has too Many Clauses

Clauses in sentences make it difficult to read. As I previously mentioned, AI tools are trained on large data sets, so they use them for generating content. When they provide content, they turn information into passive voice and add multiple clauses.

This makes AI content plagiarism-free, but it also turns into lifeless pieces of information. If you are reading an article that has too many clauses and repeats them frequently, you are reading AI-generated content.

7. Other Common Things to Detect AI Writing

Apart from the above-mentioned things, you can also note a lot of other common things to detect in AI writing. Those things are:

1. AI content is dry and emotionless.
2. AI content doesn’t always have context.
3. AI content is oftentimes boring when you read it.
4. The sentences are always lengthy, making it difficult for you to understand them properly.
5. AI content mostly hallucinates when it comes to numbers, figures, and stats.
6. The grammar isn’t always right.
7. You will see different forms of English used in content, such as UK, USA, or other versions of English.
8. The tone of AI content will not be consistent. You might start reading an article in a professional tone, and later the tone will become casual or else.
9. AI content isn’t SEO optimized at all.
10. AI content lacks personalization and human touch, which makes it unique and engaging.
11. AI content lacks creativity and vocabulary to suit different contexts and topics.

These are some of the things that you can look for in AI content.

Wrapping Up

AI tools, or generative AI, are a revolution for the marketing industry. AI tools are still in their early days, which means they can be helpful to a certain degree. As of now, we can’t use AI tools for writing blogs, or articles and expect them to give us the desired results.

If you are someone who can understand content better, you can detect AI writing without using any AI detection tool. In this article, I discuss some of the common things you can look for in AI content.

I used all of my experience and understanding and discussed it with other writers and marketers to further expand on the information. Meanwhile, if there’s something I’ve missed, highlight it in the comments section.

Before you end, also read other articles to find out more about the future of AI writing and if AI writing tools are going to replace human writers.