How to Detect Email Messages Written by ChatGPT or any AI

ChatGPT, Gemini and other AI platforms can compose human-like emails. This is how you can identify the AI.

We live in modern times that at times feel unpredictable due to technological advances in different fields. Particularly in the field of Artificial Intelligence (AI), where platforms such as ChatGPT, Gemini, Microsoft Copilot and others. It is increasingly erasing a little more than the line that distinguishes real people from entities that do not exist but behave in an absolutely convincing way.

Gone are those days when we interacted with a random chatbot that clearly felt clumsy, slow, automated and without much sense of reaction to the responses entered. Now AI has the ability to be quite convincing. And although it is capable of saving us time in tasks such as writing or organizing information. When it is used to write emails, it can generate doubts and even certain specific risks.

How do I know if the email I’m reading was written by a robot? Is it ethical to use Artificial Intelligence to write this type of messages that thousands of users receive? These are absolutely valid questions that we cannot help but ask ourselves when carefully reviewing our inbox today, in the 21st century.

Why you shouldn’t use AI to write your emails

The researcher is emphatic in warning that the correspondence generated by an AI should never be used without review or human intervention. Incidentally suggesting that content created entirely by an AI should be labeled as such. Since she considers that sending something as your own when it actually comes from ChatGPT “would be no different from plagiarizing.”

For his part, Franklin Orellana, president of computer science and data science programs at Post University, adds in the same article that AI can generate emails with syntax errors, false information or even put together pieces created for malicious purposes such as phishing. or manipulation of any kind.

Four tips to detect if the email you received was written by an AI

Thinking about potential victims, there are some alternatives to identify when we receive an email written by Artificial Intelligence. Although AI detectors exist, their effectiveness is not completely proven and in reality they tend to be redundant most of the time.

In fact, a June 2023 study titled Testing of Detection Tools for AI-Generated Text found that these tools were “neither accurate nor reliable.” But fortunately there are four basic tips that can help us:

The use of overly formal language: All AI platforms tend to generate texts using excessively formal language, which does not reflect the natural style of a human being behind the keyboard. This lack of warmth and emotion can be a telltale element.

Too uniform and/or repetitive construction of texts: ChatGPT’s Artificial Intelligence usually produces sentences and paragraphs of similar length and repetitive structure. So you only need to read a couple of paragraphs carefully to start noticing mechanical patterns in your writing.

AI uses the punctuation of a primary school child: Fortunately, emails written by AI often abuse commas and periods. So they use other more complex but natural signs in human writing, such as colons, semicolons, hyphens or even ellipses.

Hallucinations that are false: When an Artificial Intelligence articulates a text whose information is absolutely false, no matter how convincing its writing, it is known as a hallucination. This phenomenon is very recurring, especially in ChatGPT. So if what was narrated in the email is irrational, it is likely that an AI is behind it.