Algorithmic guessing misses what makes us human.
S. A. Applin, Ph.D. is an anthropologist whose research explores the domains of human agency, algorithms, AI, and automation in the context of social systems and sociability. You can find more @anthropunk.
Gmail’s new “Smart Compose” feature will auto-complete sentences in email drafts, to save users time by finishing their sentences algorithmically. It’s like auto-complete, but for entire sentences.
At face value, for busy people at a company like Google, quick ways to send out messages to help winnow down a rapidly filling in-box sounds promising. Unfortunately, Google has had a history of creating products for itself and its way of working, that when deployed en masse, can create problems in society, where people, who don’t all work at Google, communicate in a myriad of ways for many different reasons, and have different perceptions of what constitutes ‘saving time.’
It has only been a short time relative to human history that people have had network capabilities, and for the most part, we’ve adapted to using our devices—unless we are trying to use them simultaneously when doing other tasks like driving or walking. Currently, many people spend large parts of their lives connected to a network via their phones or other computers, communicating, reading, and learning, and they rapidly cycle between maintaining relationships and ties on the network and in their communities.
When we start and finish a phone call or email, we know wholly who we are and how we’ve expressed ourselves, as do those with whom we communicate. This is because our words and phrasing, tone and cadence comprise the written and spoken manifestations of our identities and personalities. When we already know each other, we have a greater context within which to frame our interactions. We may have sarcastic banter with close friends, but may refrain when communicating with a supervisor, where our humor could be misinterpreted. When we interact with someone we don’t know, we may be polite, but still ourselves in tone, for the way we phrase things is a part of how we see and interpret the world.
When our phone calls are represented by an agent as in the case with Google Duplex, or our emails are auto-completed by Smart Compose, we surrender parts of our tone and voice to be replaced by the suggestions of Google’s algorithms. This may save us time, but in doing so, it changes the way we are perceived by or perceive others. It also may change the way in which we are recognized and able maintain relationships to form ties for cooperation.
Google and other disruptive technology companies rely upon the established ways that people behave (social norms) within culture and society to use as a stable and reliable mechanism upon which to disrupt with their innovations. With Duplex and Smart Compose, Google bases its algorithms on established social norms and behaviors, which it attempts to codify and algorithmically mimic. As a result, Google can then start to deploy these patterns for mimicry as a way to "save time."
It appears that much of Google’s time-saving obsession in part draws from Taylorism, an engineering management theory developed in the early part of the 20th Century, and widely adopted throughout the engineering and management disciplines to this day. Taylorism was developed to manage time and resultant processes in mechanical manufacturing, stressing organization efficiency as a central theme. Over time, Taylorism grew and evolved to create a culture of time-saving in many industries, such as automotive manufacturing and fast food. It was initially adopted primarily in companies reliant upon manufacturing engineering. However, over time other industries and disciplines, including academia, adopted as well as taught Taylor’s methods, and they became part of the culture of the engineering and management disciplines.
As an engineering-driven enterprise company, Google has a particular definition of time saving, and as such, seems to treat the nature of communication as a necessary activity that takes too much time. In an enterprise, which is a contained environment with rules and shared goals for cooperation (and Taylorism-heavy to begin with) brief impersonal messages coordinating cooperation might even be expected. However, outside of that environment, where the world is not bound by the culture of the enterprise, agents replacing human communication for maintaining relationships and cooperative ties, doesn’t work so well.
One reason for this is that much of the world is not an engineering-driven enterprise system. At work, people may be bound by the culture of their work, which may or may not be time-saving motivated. Another reason is that people have different definitions and concepts of what it means to be efficient. For some, efficiency equates to time-saving, for others, it may be cost savings—such as spending more time to commute further for a less expensive product. The ‘regular way’ of communicating for Google, may not be a regular way of communicating for others. Furthermore, sociability and social experiences in our communities have changed as a result of mobile device use, and human behaviors and patterns have become more unpredictable, temporal, and heterogeneous—and thus, less able to be codified. People work in the gig economy, or in different types of companies or time shifts, and they constantly communicate outside the bounds of expected, predictable norms. We now relate to each other differently in physical space as we use the network space to maintain our remote ties more frequently.
As a result, outside of the domain of Google’s corporate environment, the “stable” systems that Google and others have disrupted successfully in the past, are more porous and fragmented in the present. Where the systems are cohesive, may not be in physical space at all, but in the realm of our relationships. This is where what Google is doing begins to cause damage. Google has chosen to imitate utterances such as “uh” and “um,” which it uses to convey the same personality and meaning as word choice, tone, and all else that makes us human in communication. As such, to Google, individuality is something that is treated as a stable constant, that an auto-complete email won’t threaten. But it does, because the more that people rely on Google’s brevity to impersonate or augment their communication, the more homogeneous we become to one another. Furthermore, the more bland and efficient our communication is in the ‘Google way,’ the less feeling and emotional connections we will form, and the less connected we will be with each other for cooperation.
Google’s automation can subsume all or parts of our personalities with their algorithmic suggestions. As such, the cooperation that we usually have directly with each other becomes brokered through Google. Thus, instead of directly connecting to each other in tone and voice, Google becomes the hub managing the meaning of communication. It’s like a game of algorithmic “telephone.”
Google claims it is “saving time” for people, but what it is doing is protecting itself from being disrupted by messages and emails, and by needing to call for appointments and reservations to maintain the pesky business of being in bodies which require things like haircuts and nourishment. The work ethic of the Google enterprise does not have time or patience for humans in the machine, and our chatter, physical, and emotional needs, as well as different ways of communicating are bothersome and disruptive to its way of being. Something had to be done. If Google takes the stance of “Make Google Do It,” it creates further insular behavior, keeping the enterprise and Taylorism intact towards an unattainable goal of “efficiency.’ If we “Make Google Do It,” in part, we forfeit our personalities and ability to make direct relationships and connections with each other.
The aggregate of Google’s efforts on society will be disruptive and will not yield success unless
Google learns to create flexible adaptive algorithmic technology that works more closely with the way people interact with each other. As I mentioned previously, each “time saving” device for Google generates more and more labor for people who are not part of Google’s ecosystem.
Thus, the only time that Google seems to be saving at the moment is its own.