THE BLOG

Automate This: Categorizing Humankind

08/30/2012 10:38 am ET | Updated Oct 30, 2012

All of us have a personality type. All of us, when observed by a true expert, can be categorized, tagged, and cataloged like an animal in a zoo. Are you a volatile orangutan or a docile fawn? Do you operate with guile and trickery, or are you as honest as Abe? Psychiatrists with elite skills can, just by talking with us, answer these questions with surprising swiftness. They know why we work, they know why we fight, and they know the kinds of people with whom we will get along. Diagnosticians of this caliber don't reside in most psychiatry practices; they're a rare breed. But what if their type could be imitated, reproduced, and even enhanced by a machine? What if there existed a set of algorithms that could identify our character, know our weaknesses, determine our thoughts, and predict our actions? What if we created a machine that could read our minds?

Such a development would change not only psychiatry but also all manner of commerce, customer service, and hiring practices. What if, after a business call, you could consult a bot that told you exactly how the call went and what the other side thought of your proposal? Sales calls would never be the same, bad relationships would end earlier rather than later, and negotiations would be straightforward -- until both sides employed unflappable bots to do their talking, that is.

Such technology will disrupt many of our day-to-day interactions with other people. It may sound like a far-off prospect, but algorithms have already figured many of us out, parsing our personalities and learning our motivations. We're rarely given explicit warning when a bot is observing us; it's something that's crept into our lives with little fanfare. While it may sound creepy, an algorithm that can read you as well as your spouse can be useful.

We've almost all been read by a bot, although most of us didn't realize it at the time. You likely recognize these words: "This call may be recorded or monitored for quality and training purposes." Whether it's a bank, credit card, airline, or insurance company, when we dial up customer service and hear that familiar phrase, we've allowed an algorithm to drop in for a listen. It's not unreasonable to believe that this message is merely served in case a supervisor or manager is listening in or may replay the call later. But in many cases, that's not what's happening.

What is happening is that a bot is eavesdropping on your call. It listens to you talk, assesses your personality type, and determines why you've called. In some cases, the bot relays this information to the customer service agent as they talk to you. Most incredibly, the bot tries to read your mind. How can you be most quickly and cheaply satisfied? Once the bot gets to know you -- it can figure you out in 30 seconds or less -- it will route your future calls to agents who share your personality traits. Pairing callers with the wrong agents heightens the risk of a colossal argument. Putting like-minded customers and agents together results in shorter calls, higher customer satisfaction, and bigger profits for the company.

To better understand how these bots work and the breadth of their abilities, it helps to understand their roots, which trace to NASA and, as usual, Wall Street.

Excerpted from Automate This: How Algorithms Came to Rule Our World. Published by Portfolio | Penguin. Copyright Christopher Steiner, 2012.