Did the Robot Write It? Warning Signs of AI-Generated Content

The emergence of Chat GPT and other generative AI upended the world seemingly overnight. While artificial intelligence has been around for quite some time, the sophistication of this new chatbot was seen as a real game changer.

It did not take long for people to find all kinds of uses for the tool. Workers are outsourcing their most hated and repetitive tasks to the chatbot, conveniently forgetting to tell their bosses about the robotic assistance. Students from middle school to college are enlisting generative AI to crank out term papers, sometimes resulting in a surprising improvement in their writing abilities, at least on a superficial level.

The rise of generative AI has left supervisors, college professors, and high school teachers scrambling. While plagiarism detection tools have been around for a while, they are not equipped to detect AI-assisted writing, creating a dilemma for those in charge. The good news is that AI-generated content is far from undetectable, and some telltale signs could uncover that robotic writing. Here are some signs the author of that term paper or corporate report may have had some non-human assistance.

Uniformity in Sentence Length and Structure

When humans write a prose paragraph, they tend to vary the length and style of their sentences. They may even engage in literary no-nos like run-on sentences and dangling participles.

AI-generated content, on the other hand, often displays a strange uniformity in sentence length and structure. Using tools that count the number of words in each sentence is one way to ferret out that robotic writing.

A similar red flag in AI-written content is repetitive passages, especially in bullet point or numbered lists. If you see the same phrasing repeated word for word in each item in a list, that's a dead giveaway (not the whole paragraph but a sentence or two).

Writing That Seems Too Sophisticated or Mature Given the Supposed Author's Age

Another easy giveaway for AI-assisted content is a sudden rise in the sophistication of the written content. If a student who has been struggling to write at grade level is suddenly looking like a child prodigy, something more than education may be at play.

Teachers can watch out for these sudden increases in the maturity and sophistication of their student authors. They can question the results of writing assignments and query the supposed authors on the meaning and content of what they have written. Those tests could uncover the AI cheaters and get the entire class back on track.

Outdated or Incomplete Information

The current iteration of generative AI tools has been trained on vast amounts of data, but the data sets they have been given are far from complete. The online education of the current generation of chatbots exists on a timeline, and the content they create often sounds dated or incomplete.

Specifically, AI content generators will likely not have access to the most up-to-date information and statistics, rendering the conclusions they reach inconsistent and even inaccurate. Suspicious bosses and professors may want to compare the findings of this content to the latest information; an obvious mismatch could mean the information was AI-generated.

Letter Perfect Writing

Even the best human writer is not perfect, and any sufficiently long piece of writing is likely to have a typo or two. Chatbots, on the other hand, are not generally prone to those missteps, resulting in output that is suspiciously free of grammatical, punctuation, and spelling errors.

The complete lack of typos could be due to diligent proofreading, but it could also be a warning sign of robotic assistance. At the very least, professors and workplace supervisors may want to run the results through an AI-detection tool.

As AI content generators get better and better, it will become harder and harder to tell robotic writing from its human-written equivalent. That may not be a big deal if you are playing around with those AI tools, but it could be a huge problem if you are a teacher, a supervisor, or anyone else in authority. Learning to identify the warning signs of AI-written content is a vital first step and one that will become even more important in the future.