Filed under:

Marianne Grier

Google as the Editor

Message send on mobile phone
Kitticha Polpaisal © 123RF.com

If you’re a Gmail user, you may have adopted its new Smart Compose feature. This perplexing time-saver has made me question the future of communications since it appeared in my inbox. Smart Compose uses machine learning to help users write faster, clearer emails by anticipating what they’re trying to say and suggesting text they can accept by tabbing through. It works with AI behind the scenes, scanning emails and recognizing context such as days of the week, places and other information. Smart Compose follows Google’s Smart Reply feature, which suggests three pre-written responses tailored to the email received.

One morning, especially pressed for time, I sat down to dash off a few emails. I navigated to my Gmail, opened a fresh template and inserted my two recipients’ names.  I typed an “H.” A greyed out “ello” immediately jumped out, prompting me to write “Hello” instead of the “Hi” I’d been thinking of. Sure, that works too, I thought. Gmail then suggested the names of the people in the “to” field. How clever, I thought! Except one of them was my mother, and she would think something was up if I called her by her first name instead of “Mom.”

As I worked through my email, Google suggested words and finished sentences for me. Some of the suggestions were way off. But overall, they were creepily accurate and I finished in record time. But when I read my email before sending, the tone was strange. It certainly didn’t sound like me, though the sentences I had tabbed through had been in line with what I wanted to say.

My email-writing experience made me think of Plutarch’s Ship of Theseus paradox: If the ship on which Theseus sailed has had nearly every part replaced, is it still the same ship? If Google has finished some of my words and sentences, have I still written the email? Where does the author’s voice go when emails are partially self-composing? At what point does the email I’ve signed off become something written by a bot?

Auto-correct has been a thing for ages, and most of us have had silly auto-correct moments when texting. They’re often laugh-worthy, but also make me question what’s really going on when our technology writes for us. With the Smart Compose feature, Google is no longer just the editor, highlighting spelling and grammar mistakes; it’s also stepping in as the writer, depending on the extent to which we follow its suggestions.

Have I turned off the feature? No. It makes my life easier if I’m firing off a note to our property manager or booking an appointment — cases where I feel like speed is more important than maintaining my voice. While I haven’t tested the feature out exhaustively, it seems to produce emails that are context-appropriate, correctly spelled and grammatically correct. I will continue to question how tools like this affect the way we communicate, and wonder how they might shape our communications in the future.

___

Previous post from Marianne Grier: Pink, Blue and the Singular “They”.

The Editors’ Weekly is the official blog of Editors Canada. Contact us.

6 Comments on “Google as the Editor”

  • Glenna Jenkins

    says:

    If it’s anything like Google Translate, then it’s bound to produce writing that is so bad as to be embarrassing. I would definitely very carefully read over anything Google suggests. Even in an email. Just my two cents worth on the matter.

  • Thanks for writing about your experience with Smart Compose, Marianne.

    It seems to me that if we are so rushed we can’t compose a short email without using AI, something is wrong with our priorities. We talk so much about how networking is important, especially face-to-face. We acknowledge that much of our communication is online, though; shouldn’t we at least try to make it as personal as possible? That means maintaining our voice, and also modifying it to reflect the type of connection we have with the person we’re writing to. And why not enjoy writing and reading emails? That could mean injecting humour–bots can’t do that!

  • Wendy Barron

    says:

    It’s funny because when I first encountered this feature, it felt like a godsend. And it seemed so intuitive that I wondered if it had “learned” my voice from previous emails. It goes off on a tangent sometimes, but I don’t mind still having the power to confound a mere machine. 🙂

  • Frances Peck

    says:

    Another excellent post, Marianne. I’ve been reading with horrified fascination about projects like the Japanese AI-composed novel that made it past the first round of screening in a literary contest: https://thereadersnook.com/ai-novel/

    And the computer-composed novel “1 the Road” (a play on Kerouac’s “On the Road”) which reads, says this article in the Atlantic, “as if a Google Street View car were narrating a cross-country journey to itself.”
    https://www.theatlantic.com/technology/archive/2018/10/automated-on-the-road/571345/

    As you’ve suggested, we are coming nose to nose with some devilishly complicated questions about authorship, voice, human vs. machine intelligence — stuff that forces us to question the very notion of creativity.

  • Frances Peck

    says:

    Louis Menand’s recent New Yorker article about literary hoaxes raises intriguingly similar questions about authorial voice and the (possibly elastic) notions of authenticity and identity. “Writing is a weak medium,” Menand concludes. “It has to rely on readers bringing a lot of preconceptions to the encounter, which is why it is so easily exploited.” https://www.newyorker.com/magazine/2018/12/10/literary-hoaxes-and-the-ethics-of-authorship

    • Marianne Grier

      says:

      This is on my reading list for the weekend–thank you, Frances!

Comments are closed.

To top