Google wants to finish your sentences. That's a problem.

Google recently announced it will bring its Smart Compose feature from Gmail over to Google Docs. The move will leverage the search giant's expertise in machine learning to empower consumers to streamline communication and save time, allowing them to live better, happier, healthier lives.

Ok, I'm sorry, that was awful. It probably won't make anyone happier or healthier. Google's phrase-suggesting tool — you type "looking" and it prompts "forward to it" — has from the start been a little leaden and depressing, not so much because it's bad but because it's good. I use it constantly. Because in those mundane emails about setting up appointments or going over details, it tends to suggest exactly what I need.

That doesn't mean it isn't unnerving. It's worth asking what this kind of algorithmic suggestion does to the way we speak and think. Will it make our language and writing more homogenous? Will we all talk in corporate jargon, circling back to things, hoping to pick each others' brains, signing off every time with "hope you're well!"?

Subscribe to The Week

Escape your echo chamber. Get the facts behind the news, plus analysis from multiple perspectives.

SUBSCRIBE & SAVE

Sign up for The Week's Free Newsletters

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

Sign up

The internet promised to wildly expand our cultural diversity by virtue of access to so much content. That happened — sort of. Even now, the web is full of tiny little communities with unique cultures: forums with their own vocabulary, group chats with inside jokes, successful YouTube stars you've never heard of, and thousands of little islands on Facebook, Twitter, or Instagram. At the same time, the contemporary digital is hardly a utopia of diversity. Just look at how quickly terms from African-American vernacular get co-opted by the mainstream: Everyone is "woke" now, despite that term once being a very specific black skepticism of whiteness. Niche things go viral, creators with one hit amass millions of followers, and social media is often dominated by talk of the already-famous, articles from The New York Times, and Hollywood blockbusters.

So yes, we have access to more culture than ever before thanks to the internet, but the viral nature of the internet has also increased homogeneity and sameness. Will the same thing happen to writing and speech as we rely more and more on technology to complete our sentences for us?

In her recently released book Because Internet, linguist Gretchen McCullough explains how the web is giving rise to inventive uses of language — acronyms, new ways of deploying punctuation, or just new constructions like "it me" or "lol what?"

That's true. But on the other hand, spelling is becoming more standardized, and language is changing more quickly as we are exposed to more communities more frequently. That accelerated pace means people across the world gravitate toward the same usage. For example, my family members across the globe in India are using the same slang as I — a phenomenon we would never have seen before the web.

But Google's Smart Compose isn't simply an effect of the internet's global nature, or the way in which texting or Twitter influence language. It is instead predicated on an algorithm meant to speed up and make language more efficient. The goal is not diversity or even clarity; the goal is simplicity.

Algorithms tend toward sameness. In the early 20th century, Russian literary theorist Mikhail Bakhtin wrote about two competing pressures on language: the centripetal and the centrifugal. Structures of power were centripetal forces that centralized and homogenized language, while for Bakhtin, centrifugal things like folk culture or marginalized voices — part of what he called the carnivalesque — made language diverse. Machine learning works by taking what exists and then suggesting it back to you. In other words, algorithms are centripetal in nature because they are based on feedback loops. They will inevitably suggest language or phrasing that other people are already using.

It's hard to seriously say that one new feature in one piece of writing software will fundamentally change language. But Smart Compose is part of a broader pattern in digital culture that at least beckons people toward sameness. The rise of algorithms in things like writing software isn't going to doom us, but they are a centripetal force that you can either choose to go along with or resist.

The world is too vast — and the places for people to express themselves too many — for Google to ruin language all by itself. But we are nonetheless presented with a choice about what kind of diversity we wish to see in the world, and what place tech has in either fostering that uniqueness, or simply reasserting what is the same and already exists. And without some awareness of that, we might end up "circling back" forever to one way of speaking — our answers and our writing all a product of some gentle suggestions that are nonetheless more insidious in nature.