Can Customers Tell an Email May Have Been Written Using Generative AI?

While the use of AI for emails is skyrocketing, research shows that customer opinion about it is mixed. Gartner found that 64% of customers would prefer that companies didn’t use AI for customer service. Some fear AI will provide wrong answers, while others feel that it might make it harder to get personalized support. 

That is why it is so important to understand customer expectations before using artificial intelligence to write and manage emails. In this article, we tell you all about balancing personalization with speedy email generation through AI. We also explain how Gmelius tackles this issue to provide highly contextualized, tailored drafts for every use case. 

The Million Dollar Question: Can Customers Tell an Email May Have Been Written Using Generative AI?

Today’s generative AI (Gen AI) tools like ChatGPT and Gemini can learn from giant databases of text. This allows them to accurately predict the best possible sequence of words and phrases, so they can auto-generate content—such as social media posts, emails, and even entire articles. 

Depending on the user’s input, content written by AI, such as AI-generated emails, can provide excellent results. For example, a 2024 experiment found that AI-generated emails had a 9.44% click-through rate compared to 8.46% for messages written from scratch. 

Given that AI content takes a fraction of the cost and effort to produce than traditional methods, the returns can be very high. On the flip side, AI may sound robotic, impersonal, and uncontextualized if it isn’t used correctly. Customers may pick up on these subtle cues, which might reflect poorly on the company sending AI-generated emails.

So, when you’re using an AI email assistant for Gmail, it’s important to avoid certain telltale signs that may bring down the quality of your email communications. 

20 Signs an Email May Have Been Written by Gen AI

Before using AI tools, it’s important to watch out for a few common patterns, “AI-isms,” or telltale signs that may signal the fact that an email was composed without manual intervention. 

1. Very short emails 

The AI could be lazy and respond in a single line, “Thank you! I'll get back to you shortly.” While this is suitable in some scenarios, add personal touches like the recipient's name to show that you care. 

2. No personalized details 

Most AI email assistants won't refer to previous conversations or unique details about the recipient when writing emails. After the AI-generated draft is ready, you may want to add such information manually. 

3. Empty placeholders 

When users copy-paste an AI-generated email in a rush, they might leave behind unedited placeholders—for example, “Hi! I'm excited to learn more about {company name}”. 

4. Lots of bullet points 

Most AI assistants love structure, almost obsessively, and will add bulleted and numbered lists wherever possible. Reread the draft to identify where a paragraph with commas might take more sense. 

5. Long sentences joined by commas 

AI email assistants may generate sentences with 30-50 words, with a structure that looks like this: “Cloud technology has become the backbone of modern digital transformation, enabling businesses, governments, and individuals to operate more efficiently, scale rapidly, and innovate faster than ever before.” 

Sign up for Gmelius to experience an AI email assistant with a difference.

6. A complete absence of semicolons 

Typically, AI agents tend not to use colons and semicolons when writing regular sentences and paragraphs. Clever use of punctuation while editing could make emails feel less synthetic and more human sounding. 

7. Lack of humor 

Gen AI struggles to make jokes unless explicitly instructed, and this reflects in its writing. Adding humour to email communications is best done by humans, although you can certainly work on an AI-generated draft to save time. 

8. Frequent use of title case 

Most AI agents will capitalize the first letter of each word of titles when writing listicles, bulleted lists, and sections within an email. Reread the emails to ensure that the title case is appropriate and feels natural.

9. Too many bold sections 

Highlighting certain words and phrases in bold can make emails easier to read. It also conveys the message in a stronger, more impactful manner. However, AI assistant may oevruse this technique and format arbitrary sections in bold, in a way that doesn’t always make sense.

10. Repetitive language 

In a bid to generate longer emails or sound more polite, the AI may repeat certain words and phrases. Your recipients can detect this lack of depth, and as a result, emails can become less engaging. 

11. Overuse of jargon 

The large language models (LLMs) powering AI email generators are trained on publicly available data, a lot of which is business writing. As a result, they might resort to business jargon such as “optimization” and “efficiency” without backing it up with data. 

12. Very little first or second-person perspective 

When humans write, we tend to use first and second person pronouns like “I,” “my,” “you,” and “your” mich more frequently than AI. When AI assistants use second person pronouns, it can come off as too salesy or pushy. 

13. An overexcited tone 

Typically, Gen AI avoids sounding negative and tries to convey a positive message with every email or any other piece of content. However, it can get overexcited and start using hyperbole, which sounds unnatural in email correspondence. 

14. Ineffective subject lines 

While AI subject line generators can be useful, the results are sometimes generic. For example, on your co-worker’s birthday, it might recommend a subject line like “Birthday Greetings” instead of “Many Happy Returns!” If you’re using AI for lead generation, then the subject line requires mindful proofreading. 

15. No in-depth domain understanding 

A lot of the issues we explained above is due to the fact that Gen AI doesn’t understand the nuances of a specific industry, demographic, or department. Unless it’s specifically trained on these datasets or you share industry information in the prompt, emails can sound generic. 

#FunFact: Gmelius is trained on your industry-specific communications. Try it with a free trial.

16. Complex words when simple will do 

While this isn’t unique to AI (human writers do it too), the content may use complex, difficult sounding words instead of simpler alternatives. Common examples include “meticulous” instead of “careful,” and “pivotal” instead of “important,” especially when they are overused. 

17. Overly formal 

AI-written emails can sound too formal, almost cold, unless the LLM is properly trained or instructed. This is because most AI generators draw from a database of pureky business writing, and unlike humans, they can’t interpolate humor and storytelling skills, with a business context. 

18. Absence of white space 

The structure of AI generated emails can also be a telltale sign to recipients. They are likely to notice dense, long paragraphs one after the other instead of a few short sentences followed by the core message. 

19. Highly formulaic greetings 

Most email recipients are tired of seeing messages that start with “I hope you are well” or “I hope this email finds you well.” This can be a telltale sign of AI writing that hasn’t been proofread by humans. 

20. Misplaced emojis 

In order to sound more positive and upbeat, AI generators can overuse emojis, particularly the graphics for rocketship, fire, and red alert. These can be distracting in an email, and take away from the overall message you’re trying to convey. 

How to Use AI for Personalized Emails that Don’t Sound Generic

While AI for emails can save you hours of time and manual effort, it’s also important to recognize its shortcomings. By keeping an eye out for these 20 telltale signs, you can make small but significant tweaks that make messages exponentially more powerful—without sacrificing the efficiency gains from AI. `

If you’re using a prompt-based interface, include your desired email structure in the prompt. This will prevent the AI from generating blocks of similar sounding paragraphs. Also, include contextual data like customer history in your instructions. This allows the AI to combine specific information with its general training to produce better results. 

Finally, avoid reusing the exact same prompt repeatedly as this can cause monotonous results. Change small details like the customer context, the greeting/salutation, or a reference message with every prompt. This helps the Gen AI model to keep learning and stay on its toes. 

AI Efficiency Meets Personalized Training and Tailored Automation: That’s Gmelius! 

Gmelius AI agents help emails sound less generic and ensure your own voice always shines through. It is trained on the thousands of emails you send, so the agent can learn your communication style and preferences. 

What’s more the training is reinforced every week using new data from your inbox, so it can adapt and become even smarter.

To make the AI more knowledgeable, users have the option of training agents from their website, help center, wiki pages, and other sources. This ensures that messages are informative and highly specific to the subject matter.

Another challenge when using AI for writing emails is knowing which emails need urgent attention. The Gmelius AI agent can differentiate between messages that merit a response (for example, a customer query) and those that don't (for example, promotional emails). 

Further, when it doesn't have enough context, the agent leaves notes explaining why it couldn't draft a reply. This avoids AI “hallucination” and guarantees that you never send out a generic or out-of-context email. 

Conclusion

If you’re wondering, can “customers tell if an email is generated using AI,” the answer is—mostly—yes. Most Gen AI models include “signatures” or telltale signs in their writing, which can make emails seem generic and lacking in depth.

That’s why Gmelius is building a proactive AI agent that constantly improves with weekly reinforcement learning based on new datasets. We enable greater personalization and a smoother experience for email writers, while saving you 1 hour of work every day. 

‍Sign up for Gmelius to try the next generation of AI-powered email writing.