Your Email Box May Explode Thanks to Generative AI in Gmail

Gmail's new generative AI feature promises to bring every office worker's nightmares to life. 

A lot of office work is busywork, and dealing with pointless emails is already a big waste of time. You're probably sick of your co-workers delegating tasks to you instead of just doing them, and now they can pretty much automate their work-shirking antics with Gmail and Google Docs' new generative AI features. So, is this a nightmare or a productivity life raft?

"There will certainly be many unintended consequences of generative AI. And the increased volume of email could certainly be one," Tom Taulli, the author of the book "AI Basics," told Lifewire via email. "Although, ironically enough, generative AI could help deal with this—to some extent. Summarization is a key feature. So it may become more common to do this with incoming emails."

Generative AI: The Inbox Hero

The new generative AI features for Google Docs and Gmail are currently being tested as part of Google Workspace. The idea is that you give the AI a text prompt, and it goes from there, writing emails, creating reports, and so on. And it's not just text, either. It can also create images for your presentations and "bring your creative vision to life with auto-generated images, audio, and video in Slides," says vice president of product for Google Workspace Johanna Voolich Wright in a blog post

The potential is huge. All that annoying corporate busywork is exactly the kind of thing that a computer should do, not a human. The dream is that, instead of spending days writing that presentation about the project you've been coding, you could just tell the Workspace AI to draft a presentation for you and get on with some more coding.

But as we have seen, AI isn't all that when it comes to accuracy and truth. A conscientious person would have to thoroughly read, vet, and edit the AI-generated text for errors. The harried corporate executive, however, might just give it a once-over for tone and send it on its way.

The Loquacious AI

As a journalist and erstwhile editor, I have sometimes had the pleasure of editing less-than-stellar copy. In those cases, it would literally be quicker, easier, and give a superior result if I were just to ditch the original and write it from scratch. Which is to say, editing an AI's homework might end up being more work. And, unlike a human to whom you can give feedback and guidelines for future articles, the AI will presumably be just as enigmatic every single time. 

And then we get to email, the point of this article. If you're already swamped with coworkers shunting a task onto you, then AI is about to make that rain of pointless work into a tropical storm. Here's how easy it is to generate a few paragraphs of full buzzword-compliant corporate-ese. The following still from Google's intro video show what happens when you just type "I'm on it" in reply to an email. 

The original prompt, "I'm on it," would have been far superior. 

"So far, most AI-generated text has been put to use on marketing and outreach tasks rather than internal communications. The issue with ChatGPT for these kinds of emails is that what it really struggles with is specificity and detail, and these are the most important parts of any worthwhile work email," Ben Michael, attorney at Michael & Associates, told Lifewire via email.

There's a twist, though. One of the AI's features is to summarize an email thread. You can probably see where this is going. Instead of reading through a stack of mixed-origin (AI and human) emails, you can just tell the AI to summarize the thread. 

What a beautiful irony it would be if the 'bot summarized the above flapdoodle into "I'm on it."

The Legal Complications of AI

There are other downsides. Like any tool, this can be used for both good and evil. 

"Generative AI makes it easier for hackers to create phishing emails. This is especially for those who do not write English particularly well," says Taulli.

And although it's beyond the scope of this article, there are concerns about copyright, plagiarism, and other legal landmines when you are using an AI to generate words and images because AIs are trained on vast troves of existing human-created—and most likely copyrighted—material.

What if the AI plagiarizes an existing text and passes it off as official company business? Or if it generates a report, including recommendations that turn out to be false and even dangerous? Who is liable? We don't know, but we definitely won't be asking an AI for the answer.