Have you ever stopped to think if the piece of content that you’re reading was written by Artificial Intelligence (AI)? We live in a world where AI is becoming increasingly prevalent, and it can be hard to know when its influence begins and ends. In this article, we will uncover how you can identify whether or not something has been written by an AI-powered program.
We have all seen stories about robots taking over our jobs and replacing human workers with their efficient algorithms - but what happens if they start writing articles? AI has become so advanced that sometimes it can be difficult to tell the difference between something written by a machine versus a person. Learning how to spot these pieces of content is essential for maintaining accuracy in today's digital age.
So, if you're curious to see if an AI system wrote your favorite blog post or news story, then keep on reading! We'll look at some key indicators that will help you discern which pieces were composed by artificial intelligence and which weren't. With this knowledge in hand, you'll be able to make sure that your sources are reliable every time.
AI Writing Tools Aren't The Enemy
This is a topic that can be very emotional to many people and the power that tools like this offer can be a very addictive drug. Simply put though, AI writing tools by themselves are no different than using Gramarly to edit your human writing by suggesting improved sentences or word processor instead of dictating something to your secretary the way one might have 50 years ago. Instead, the challenge will be in how these tools are used and what they are used to create.
As an example, if a student uses a tool like ChatGPT to create an essay instead of writing it himself, that's no different than using cliffnotes, buying copies of old exam questions, or using an essay writer. Technology has changed, but the intent to deceive has not.
However, if you hire a content marketing company to produce top tier talent that draws search traffic and attracts links - should it matter if that team uses tools to make them more productive, more creative, and better optimized? It shouldn't. You should be happy to get better quality content than was possible without using technology!
This is clearly an area that will evolve over time, and it is going to be a very exciting future that we get to create together.
AI Writing Tools Aren't All The Same Either!
While most people reading this are probably familiar with ChatGPT, OpenAI, or GPT-4 (all derivatives of the same basically language model), the reality is that at the base AI tools produced by the OpenAI company are then adapted through the use of an API and other companies have created unique layers on top of it. For instance, while Jasper and ZimmWriter are both drawing from a GPT-3 and GPT-4 API, the output is significantly different and even more diverse compared to the output you might get from a free AI writing tool like using ChatGPT directly.
As a result, it becomes increasingly difficult to identify any specific output as being conclusively AI written vs 100% from a human writer.
Some Ways That You Can Identify If AI Wrote The Article
There are many clues that can lead you to believe that AI may have written an article but the reality is that like with everything out there, things are not perfect. In fact, the entire concept of "outing" a piece of content regardless of quality, tone, and enjoyability is really beyond the point of consuming written content. At the end of the day, each writer has his or her own style and for the past several years we have existed in a world where AI tools and Human creativity have become blended. This will only increase as the tools become better.
With that being said, there are some clear indicators that you can use to identify poorly written content that was produced using AI writing tools. Let's take a look at some of the most obvious ones ...
Contextual and Relationship Errors
Years ago, I had an article commissioned about the history of fried chicken - in theory from a human writer commissioned through TextBroker. However, it was clear that it had been written by AI since it began talking about breakfast and fried eggs being popular in places like Kentucky.
Why yes - people in Kentucky do make EXCELLENT breakfasts consisting of fried eggs, sausage, and pancakes ... maybe even with some delicious bourbon maple syrup - the AI tool had made a false contextual jump from "fried chicken" (presumably Kentucky Fried Chicken) to "fried eggs" since well chickens come from eggs.
A human writer, especially a native English writer, would never have made that connection because most humans understand the context of eggs vs chicken even if they are essentially the same animal and both can be enjoyed fried.
This is probably the number one thing that editors as well as clients that are reviewing AI articles from their content teams should look for since it is by far the easiest to identify by simply reading the article.
Typos, Errors, Inconsistencies, Poor Punctuation, And Bad Facts
AI tools, notably GPT-3 and GPT-4 tools are notorious for creating facts that are blatant lies. Open AI refers to these as hallucinations and the reasoning behind why they are a phenomenon is more complex than we can cover here but the prevalence is notable to the point where I personally assume every quote, fact, or reference is falso.
However, there are also other elements beyond these bad facts that you need to watch for. If the article is riddled with typos, errors and inconsistencies - or even worse, has bad facts listed - it can often indicate that this piece of content was not written by a human. Similarly, strange grammar mistakes and incorrect punctuation are red flags for artificial intelligence writing.
When reading through the content, take note if any parts feel robotic or unnatural in their structure. This disjointedness in language might reflect how the AI algorithms were programmed to write, thus making it much easier to spot than if a real person had crafted it.
Additionally, editors should pay attention to whether certain words appear too many times or in wrong parts of the copy. This is a combination of factors including that AIs will likely repeat them more frequently than humans would due to its lack of understanding context, as well as tools designed to produce "SEO Optimized" content may try to insert words without looking for natural variations.
The tell-tale sign of AI written content is its absence of consistency. It's like a tangled thread, weaving in and out of topics without cohesion or structure. This is as much an indicator that an AI tool was used to generate a "One Click Article" or a prompt was used to "Write me an essay", as it is that an AI tool was used in the first place. Advanced AI writing tools will require that the author create a well-structured brief with topics, keywords, even URLs to reference.
To me personally - even just simple human input like this immediately separates an article or blog post generated using a tool like ZimmWriter from one where a college kid is trying to "cheat" on his essay or a content writer is trying to fool his client by simply using free tools like ChatGPT.
When you see it, the result is obvious. It's as if the narrative has been broken apart and put back together again from an entirely different perspective. To determine whether something was written by AI or not, one should look for these key signs:
- The lack of cohesiveness between ideas
- Grammatical errors that appear to be generated randomly
- Unnatural sentence structures which don't follow any established writing conventions
- The repetition of keywords without any meaningful context.
Consistency in this context can be both bad and good though. For instance, while variations to the extreme are negative - articles that are too consistent in the number of sentences per paragraph or paragraphs per topic would also be considered indicators.
Lack Of Authenticity And Other Human Elements
Lack of authenticity and human elements can be another strong indicator that an article was written by AI. However, it is not a guaranteed way to identify articles that may be generated by AI tools since it is relatively easy to mask this by using slang or literary expressions. These additions though have the added benefit of making the writing more approachable and easier for the reader to connect with what the author is trying to convey. In this regard, those "tricks" may not be a negative thing. However, articles or essays that are completely devoid of expressive elements can be suspecious.
AI-generated articles often lack the nuance, subtlety, and complexity that are associated with well-crafted writing. AI-generated articles are typically written in a simplistic, formulaic style that lacks personality or emotion. They also often include the same words and phrases throughout the article, making them sound repetitive and robotic. Additionally, AI-generated articles may lack creativity, as they often rely on templates and pre-programmed algorithms to generate the content. As a result, AI-generated articles can easily be identified as lacking authenticity and human elements.
However, the biggest "Human" element that will be missing in any AI-generated content is that it truly lacks expertise. While the AI can be trained to write, "I am an expert" or recite facts and figures (or just make them up completely), articles that lack complex thought or personal elements that are identifiable from the author's history will be easy to identify as being potentially AI written.
When assessing if the content was written by AI, one characteristic to look for is repetitive phrasing. This often happens when AI generates text because the same words are used multiple times in a row or repeatedly throughout the piece of content. For instance, if you noticed phrases like ‘AI writing’, ‘content written by AI’ and ‘written by AI’ repeatedly in an article, it would likely indicate that the article was composed using AI software. However, you will notice in this article that I added many other variations of that phrase as would be natural when something is written by a human.
In summary, looking out for phrases being repeated as well as overly complex sentences can help determine whether content has been created by AI or not.
Wrong Forms Of English - British Vs American
When it comes to English, there are two major forms - British and American. While the differences may be subtle, they can have a huge impact on how an audience receives content. Here are four key areas of difference between British and American English:
- Vocabulary - Different words tend to be used for certain objects in each dialect. For example, a ‘lift’ in Britain is what Americans call an ‘elevator’; likewise, a 'biscuit' in America refers to something different than a 'cookie' in Britain.
- Grammar - There are slight grammatical variations between British and American English as well. An important one pertains to collective nouns such as “team” or “committee” which take singular verb forms in American English while taking plural verb forms in British English.
- Spelling - The spelling of some words is also different depending on whether it follows the British or American form of English. Examples include colour (British) vs color (American), organise (British) vs organize (American).
- Punctuation- Even punctuation marks differ slightly with periods following abbreviations being placed outside quotation marks in British usage but inside quotation marks in US usage .
Human writers will general speak (and write) from one of the two styles. AI sometimes will mix these things together or use the wrong format. When it is used properly but placed improperly in a blog or submitted as a college essay for instance, that is a very good indicator that the content either came from an AI writing tool or an essay writing service. Regardless of the source, the fact remains that a human failed to edit it and the person claiming credit for creating the piece may not have actually written it.
Have you ever read a piece of content and felt like something was off? That it wasn't quite right, but you couldn't put your finger on why? The text you were reading was likely created by an AI tool. To confirm this theory, look out for formatting anomalies. AI tools often have difficulty with grammar, syntax, and punctuation - so if you spot any errors or mistakes in these areas, then it's a good indication that the content is computer-generated.
AI Writing Is REALLY GOOD - Human Editing Is Still Required
It can be hard to tell if a piece of content was written by AI, but some signs give it away. Looking for consistency and repetitive phrasing is an easy way to spot something machine-written, typos, errors, inconsistencies, bad facts, and formatting anomalies should also raise suspicions.
However, these are all things that can be tweaked when a human editor is handed the output to revise and update. When you combine the two (Human + AI), the result is an achievement that should be celebrated, not feared.
While checking for AI writing, the most important thing to remember is not to assume the worst at first glance. After all, these tools have advanced significantly over the years and may have been used in good faith by their creators. It’s worth taking a few moments before jumping to conclusions about what you read or hear online since many news reporters simply don't have all the facts and lack the experience necessary to properly understand what's going on.
For me personally, I've been using AI writing tools for several years now - both as assistants
At the end of the day, though, I believe that artificial intelligence has its place – even within writing. As long as writers remain vigilant and exercise critical thinking skills with anything they come across on the internet, AI shouldn't pose any serious threat. With time and effort put into research and analysis, anyone can easily identify machine-written pieces of content without too much trouble!