A friend and fellow writer I hadn’t heard from for a while sent me an email the other day. She wanted some advice, and I readily offered it. Then she wrote me a nice response–the sort of casually witty, stupidly impressive little thing that no one I know could have written but her.
And at the bottom of her thoughtful email, Google offered me three quick responses in the iOS Gmail app:
Sounds good to me!
With a tap, I could auto-fill my response to her. Just like on Google’s Allo Messenger or Apple’s iMessage, I could let the software respond to an email or text not with a unique response, but a quick tap to say something like “haha,” “lol,” or “talk later.”
Sometimes, it’s tempting! Sometimes–especially dealing with work emails in the off hours–I have hit those quick response buttons. But I also know that it’s very bad for me, and it’s bad for us, as a collective, creative species. Today, we face one of the promises of Silicon Valley: to quantify us. To average our individuality out. To train a model on all of those collective averages, creating the ultimate average response. To provide the convenience of never speaking uniquely again.
Perhaps the most bothersome examples are Google’s AI-powered sketching apps. One, Quick Draw!, invites you to draw a beard or cake or any other number of things, while a neural network tries to guess what you’re doodling. Recently, Google showed off the results of the app: 50,000,000 sketches that illustrate how we are not so much unique snowflakes as a collective of people who will all, independently, draw exactly the same rendition of a snowman.
Google wouldn’t tell Co.Design exactly what it’s doing with the data itself. We don’t know how Google will learn from all of this free human labor and, presumably, serve it back to us as a product, the way it uses its vast troves of data to improve the accuracy of many Google services. But Quick, Draw!, and its matrix of countless identical sketches, reinforce a Silicon Valley stereotype: that human behavior is just so average and tedious. So give an AI the keys to your car, or your Creative Cloud account, and the machine will do a better job while saving you some time.
Take Google’s other drawing experiment, Autodraw. Autodraw takes Quick, Draw!’s premise even further. You scribble a sketch. Google analyzes it, works out what you’re poorly attempting to illustrate, and replaces it with a generic piece of clipart. It’s pretty much Adobe with auto-complete. Progress!
Now I know what you may be thinking: “I can’t draw well, so Autodraw sounds great to me!” Or maybe, “If I was going to type ‘lol’ anyway, who cares if Google does it instead?”
Well, future generations of thinking humans care. Consider how scientists found that the average literate person’s vocabulary has shrunk over the last two centuries, after analyzing unique words used in books since 1800. In exchange for awesome technologies like television, text messaging, and an app called “Yo” that let you type a single word (and raised $1.5 million for it), we slowly handed over the ways we can express how we feel and what we think.
Some of this really was progress. Emails and texts connect us easily and inexpensively. Platforms like Instagram and Snapchat allow us to share more pictures than ever before, which is great! The problem is with systems that actively presuppose, assume, and auto-complete what was once self expression. Text messaging may have ushered in a collective slang, but even that is now being molded into a 12-response vocabulary that the machine selects for us. Is this supposed convenience–these smooth averages of human communication–all that Google will give us in return for the free data we contribute by using its apps and playing its cute AI games? And what happens to the rest of that data? Does it become some proprietary asset at the core of Google, that only Google can profit from?
As to the promise of personalization based on your data, even people within Google have told me that it uses algorithms in a mostly aggregated way. For instance, you make a small data donation, like sharing your car’s location with Google Maps, and as a result you get a map of traffic in real time around your city. You benefit from the aggregated data, but it’s not necessarily tailored to you. The map is the same for everyone. Your route through it is the individual bit.
When we communicate with language, we are not always looking for an optimal route through traffic. In fact, it’s the side trips–the jokes, the phrasing, or the specific motion of the pen–that make a person unique. Once you get all the commas and the periods in the right places, it’s the subtle differences in the way we communicate that makes language so valuable.
What really scares me about the rise of aggregated, averaged, auto-completed culture isn’t just that I feel it chipping away at my own vocabulary, but I fear it will will teach young people how to speak via an anonymous algorithm before they can develop their own splendid, flawed voices, before they can invent new words, and new forms of self-expression, that will enrich our culture and progress as a society.
It sounds dramatic, doesn’t it? Google is coming for our artists! But I want you to think of your favorite author or artist who bucked social norms to herald a new era of human expression and meaning. Now imagine that, instead of creating the most impactful work of their career, they phoned it in that afternoon with an auto-completed sentiment.
Crazy, right? But what can you do in the era of these machines? Ball up all your angst and tell me which of these options best applies:
Sounds good to me!