This article is part of Finding the Words, a newsletter that delivers practical insights on the day’s issues.

This week’s article was AI-generated. Or was it?

It’s an important question for you to answer as AI-generated writing is quickly becoming indistinguishable from human writing.

• Did your employee write that report or turn to AI for it? 

• Did your student write that poem, or did an AI app write it for them?

• Did your top job applicant write that cover letter, or did AI?

• Is that email requesting credit card information real or an AI-generated scam?

How to tell? 

If you’ve started to wonder for yourself, the good news is there are several clear signs of AI-generated text—and you can be trained to spot them.

A team of researchers at the University of Pennsylvania School of Engineering and Applied Science found that people can learn to spot the difference between machine-generated and human-written text—and discern the reliability of sources in the process.

The researchers built a game called Real or Fake Text? to test how many sentences a computer can generate before a human can regularly spot AI text. They found that people get gradually better over time. It takes practice, and the learning models are constantly evolving, but with some practice, you can better spot AI text, too.

In addition to building up a muscle for AI detection through practice, there are a few other telltale signs that writing may be AI-generated:

• Buzzwords. AI tools regularly use jargon to fill gaps in knowledge. If the content is cluttered with complex, technical terminology, it may be AI-generated.

• Missing Context. AI models lack real-world understanding of issues, so if you’re reading a document that seems unable to provide context or is missing a central point, it may be AI-generated.

• Forced Expressions. AI tries to use idioms and slang to sound human but tends to sound just a bit…off. If you see idioms or language that feel forced or out of place, AI may be at work.

• Lacking Opinion. AI is based on algorithms, not reason. So, if you’re reading an argument that sounds colder and more calculated than compelling, with limited reasoning, it may be AI.

• Little Personality. Human-generated content often includes anecdotes, personal stories, and real-life examples. However, AI-generated text usually feels basic and lacking in personal touches, making it feel generic and unemotional.

• Patterns are Present. AI systems are based on the information they are fed, and they rely on complex algorithms that often generate text with repeating words or sentence patterns. Be aware of repeating phrases, too many emojis in social media content, or an overuse of alliteration.

• Perfect Grammar. Even the best human writers make mistakes. While all writers strive for perfection, only AI models can achieve perfect grammar every time. So, if the writing feels too perfect or your employee seems to have drastically improved as a writer overnight, you may ask if they’re turning to AI tools for some help.

As I’ve shared in this column before, AI is most powerful when used to augment and enhance human intelligence rather than replace it. While the use of AI in communication moving forward is inevitable, it is only human to have a brain, a heart, and a conscience. If you want to distinguish between real and fake, look for the human between the words.

Oh, and if you were still wondering, this article was NOT AI-generated, though I did consult my digital twin to pull up a summary of my past articles on AI.

This post is part of the Finding The Words column, a series published every Wednesday that delivers a dose of communication insights direct to your inbox. If you like what you read, we hope you’ll subscribe to ensure you receive this each week.