The Most Human Human

Brian Christian’s The Most Human Human: What Artificial Intelligence Teaches Us About Being Alive is a 2012 nonfiction book about Christian’s personal journey to learn what separates a human from a machine. In an age of ever-more-advanced AI, he explores what computers are and are not capable of, embarking on a quest to win the title of the “Most Human Human” in the annual Loebner Prize competition. Christian is a writer with degrees in computer science, philosophy, and poetry. His work has appeared in The Atlantic, Wired, and other publications.

As Christian explains, the book’s genesis lies in the annual Loebner Prize competition, which puts AI to the Turing Test. This test, named for early computer scientist Alan Turing, is based on his prediction that by the year 2000, computers would be able to trick 30 percent of human judges into thinking they were talking to a person. So, in the Loebner Prize competition, artificial intelligence programs and human confederates compete to answer questions and hold a conversation with a panel of judges, who cannot see the contestants. The judges then vote on which participants were human and which were AI. The AI that best fools the judges wins the Most Human Computer Award. But there is an award for the most convincing confederate as well: the Most Human Human.

No machine has ever fully passed the Turing Test. But in 2008, the best AI program in the competition is one vote short of passing—the closest any computer has ever come. Subsequently, Christian is chosen as a human confederate for the 2009 test. He decides that he will vie for the title of Most Human Human. But to do so, he needs to understand how to give the most “human” answers. He spends the year speaking to experts across all different fields to determine what separates humanity from AI, what qualities we have that a machine cannot have.



Much of the book focuses on how people converse with each other, and the ways that AI programs, such as chatbots, try to mimic those conversations. First, Christian considers the form and content of our speech, the idiosyncrasies that make us sound like ourselves, and the ways technology can change that.

He discusses the rise of chatbots like the Cleverbot of the early 2000s. AI programs have become good at mimicking certain aspects of human speech, such as how to change topics, make non sequitur statements, and even joke. For a time, Cleverbot was so good at imitating human speech, some users suspected it was all a hoax and there was a human on the other end of the chat log. In fact, it was learning and storing the “correct” responses to certain snippets of conversation, such as “Hello,” from the input of hundreds of thousands of users over time.

In searching for the spark of what makes us human, Christian realizes that perhaps the near-victory of machines in the 2008 Loebner Prize competition isn’t that machines are becoming more human, it’s that we’re becoming more like machines. Call center workers, for example, are trained to use certain canned responses to callers, responses that can make them sound just like an automated voice menu.



The rise of technology can also limit creativity: for example, Facebook removed users’ ability to write out their favorite activities, replacing it with a drop-down menu offering a list of preselected choices. In another example, the autocorrect feature tends to overcorrect, often assuming that some piece of wordplay is a typo or unnecessarily changing certain words in a conversation, distorting the meaning.

Christian also delves into the famous game of chess between human master Garry Kasparov and computer Deep Blue. The computer won; it was a seminal moment for artificial intelligence. But he discovers that Kasparov believes the game didn’t really count: he lost not because the computer was truly better at chess, but because he made a simple blunder early in the game and did not recover. The entire match, Kasparov says, was played within the realm of well-known moves to begin and end a game of chess, moves master chess players all know and refer to as “the book.” In his eyes, a true victory would be one that went outside of the book, a game whose victory relied on strategy and true understanding of the game’s logic rather than memorization of the moves in the book.

Finally, the Loebner Prize competition comes around. This time, the machines lose ground: not a single judge offers a vote to an AI program. The humans win, and Christian gets his wish: he is awarded the prize of Most Human Human.



Christian is not a technophobe: he appreciates what new technologies can do for us. But he is also eager to preserve the best of our humanity, the nuances of conversations. He believes that perhaps what separates us is curiosity, the constant desire to learn and grow. A machine cannot really be curious. It knows what it is taught. Christian calls for his audience to reject complacency and to be curious and engaged in the world around them.

The Most Human Human was well received upon publication. It became a Wall Street Journal bestseller and was named a New Yorker Favorite Book of 2011. A review in the New Statesman called the book “lively and thought-stirring,” and The New York Times declared that Christian “deserved his title” of Most Human Human.