9 ways to know whether an email or LinkedIn message has been generated by AI

Cover Image for 9 ways to know whether an email or LinkedIn message has been generated by AI

Why might you care about detecting AI-generated messages?

Well, why wouldn't you? It's fun to know! We all secretly love a bit of detective work. And none of us want to be duped by an artificially intelligent machine.

But on a serious note, perhaps either of the following applies to you:

  1. You're a busy person and your email and LinkedIn inboxes often receive messages from people you don't know. Some of those messages are useful, but most just seem like spam. It would be handy to be able to very quickly distinguish the useful from the spam.

  2. You run outreach campaigns and you know automated outreach doesn't achieve the response rates of manual outreach. You find that automated outreach loses authenticity and runs into the risk of seeming like spam.

If used incorrectly, AI-powered message generation can quickly achieve the opposite effect than intended when using it for outreach messages. It turns people away, rather than cultivating their interest... which is a real shame if the message is talking about something that should or could be of genuine interest to the recipient.

More and more outreach messages are being automated - so how can we avoid wasting our time on spam?

Whether we like it or not, every aspect of outreach is increasingly becoming automated - from deriving a list of targets, to executing the optimal steps in a campaign, to fleshing out the details of the messages sent by that campaign.

Particularly with the release of ChatGPT for general purpose use, messages in particular are becoming more and more automated.

But there is good news! At Koneksi, we have spent many, many hours prompting GPT and analysing its responses. So we have an experienced perspective on the things that it does regularly - and in particular, the things it does that we as humans wouldn't do.

GPT effectively leaves some fingerprints in its messages, and we know exactly how to spot them.

So in the hope that it helps you to either 1) sort the real, useful messages from the spam in your inbox or 2) choose to use automation in a much more humanlike way, we've compiled a list of those fingerprints for you.

Hope it helps!

Primary Fingerprints - these fingerprints are absolutely everywhere

This first group of fingerprints are the biggest tells. These are the hallmarks of GPT-powered copy. These fingerprints are individual words which are used so commonly that when you see them, it almost guarantees the outreach message was written by GPT.

  • Any version of the word: 'impressed'

For example: "...and I was impressed by your cloud-native PMS, empowering hotel staff, guests, and operators."

  • Any version of the word: 'fascinated'

For example: "...and I was fascinated by its unique approach to detecting fraud."

  • Any version of the word: 'intriguing'

For example: "...and I was intrigued by your mission to empower people, optimize processes, and maximize technology."

  • Any version of the word: 'innovative'

For example: "...and I was impressed by your innovative AI-powered solutions."

Secondary Fingerprints - these fingerprints are less common, but in combination are a sure indicator

This second group of fingerprints are individual words used by GPT less frequently, but in a way we wouldn't typically use them as humans. And in combination with any other fingerprints these words can still give you a pretty strong indication that the message was written by GPT.

  • 'delve into'

For example: "As I delved into your website, I noticed some interesting industry trends closely relevant to you."

  • 'eager to'

For example: "I'm eager to connect and learn more."

  • 'caught my attention'

For example: "...your innovative approach to automating manual checklists and eliminating compliance risks caught my attention."

Tertiary Fingerprints - the voice of the message

This final group of fingerprints are more about tone and less about individual words.

  • Often, the tone of the GPT-generated message is too excitable. It uses superlative adjectives, or it emphasizes words with adverbs in an inauthentic way.

For example: "...I was impressed with your company's expertise in Microsoft Dynamics and NetSuite. Excited to connect and learn more."

Or another example: "Your company has the potential to revolutionize go-to-market teams."

  • Also, the approach used by GPT-generated messages can often be too complimentary - which ultimately comes across as sycophantic.

For example: "...I was impressed by how your software empowers enterprises to achieve a state of true data management."

What conclusions can we draw?

So what's the moral of the story? Is GPT over-hyped? Will it ever be able to sound human? Should society collectively make a decision to stop using it for generative outreach messaging?

Well, GPT is not going away. In fact, it's only going to get more prominent and more powerful.

So our best chance is to learn how to work with it, and how to use it in a way that returns results we're happy with.

If AI-generated messages can be as human as you or me, the messages we all receive are going to be much more relevant - because they're going to be able to get the right points across to the right people, in the right way.

And once these messages are really sounding like you or me, there is no limit to the amount you can scale your outreach. Which means there is also a huge uplift to be had in terms of engagement rates - and subsequently lead conversion.

At Koneksi, actually we really do believe in the power of GPT - and we can see that generative messaging has incredibly high potential. But only when GPT is harnessed appropriately.

So that's why we've made it our mission: harness the potential of generative natural language by knowing exactly how to make it as humanlike as real humans. Our messages have no fingerprints. They just have real, useful information, articulated in an intentionally humanlike way, for real people to read.

Koneksi's personalised messages are the market leading option for scaled, human-like, high-conversion outreach. Head to our homepage to find out more!

Charlie
Charlie

Read more

From the Blog

We share more about sales tech, trends, tips and behind the scenes in the Koneksi Blog.

The people building whole software companies on their own - how they do it, and why.

The people building whole software companies on their own - how they do it, and why.

Discussing the concept of indie hacking, and how it compares to venture capital as a method to grow a software business.

Read More
Announcing the launch of our new managed services suite!

Announcing the launch of our new managed services suite!

We've listened to community feedback and in response we've decided to introduce a series of official managed services - so that we can run your LinkedIn DM campaigns for you.

Read More
We launched on Product Hunt for the first time and here's what happened...

We launched on Product Hunt for the first time and here's what happened...

We took our ROI Calculator to Product Hunt on Saturday and this is everything we did, experienced and learned.

Read More

Join our community of innovators

Get the Tools and Insights That Set You Apart

Be first to hear about new insights we uncover, new free tools we make, and get updates on the latest ways to use AI in as human a way as possible.

Your subscription could not be saved. Please try again.
Your subscription has been successful.

🔥 Join our passionate AI community 🔥

Get tips on new tools, new insights, and the most human-like AI.

By subscribing, you agree that we can transfer your information to Brevo, our marketing platform, in accordance with their terms of use.