The Most Human Human by Brian Christian

This book

Synopsis
Reviews
Product details
Interview

About Brian Christian

» View larger jacket image

The Most Human Human

What Artificial Intelligence Teaches Us About Being Alive

» Brian Christian

Viking Adult
Hardback : 05 May 2011

£18.99

» Buy now

Synopsis

For the first time in history, we are interacting with computers so sophisticated that we think they're human beings. This is a remarkable feat of human ingenuity, but what does it say about our humanity? Are we really no better at being human than the machines we've created?

Computers are now so adept at behaving like humans that they are on the brink of passing the Turing Test, the widely accepted threshold at which a machine can be said to be 'thinking' or 'intelligent'. In this brilliantly witty and inspiring investigation, Brian Christian explores first-hand the urgent moral and practical implications of this remarkable development. And in an era when so much digital communication is metaphorically - but also quite literally - a Turing Test, he explores how to be the most human humans that we can be.

Drawing on science, philosophy, literature and the arts, and touching on aspects of life as diverse as language, work, school, chess, speed-dating, art, video games, psychiatry and the law, The Most Human Human shows that far from being a threat to our humanity, computers provide a better means than ever before of understanding what it is.

» Back to top

Reviews

» Submit a review

» Back to top

Interview

1. In 2009, you were awarded a remarkable prize for being 'The Most Human Human'. What does this title, and the title of the book, actually mean?

The title comes from the Loebner Prize competition, one of the artificial intelligence community’s most anticipated and controversial annual events. The contest is what’s called a Turing test; the name comes from mathematician and founder of computer science Alan Turing, who in the 1950s, as the computer was just being invented, was already considering the philosophical implications of these new machines. Specifically, can machines think? Would it be possible to create a machine someday that could? And if so, how would we know?

Turing put theorizing aside and instead proposed a practical test: assemble a panel of scientists to have five-minute long text conversations via computer terminals, sending and receiving text messages. The catch is that the judges don’t know whether the messages appearing on their screens are coming from real people, or from computer programs pretending to be real people. Turing famously predicted that by the year 2000, computer software would be fooling the judges 30% of the time, and that as a result, we will “speak of machines thinking without expecting to be contradicted.”

The annual Loebner Prize held since the early ‘90s has put Turing’s theory to the test, and although this famous millennial prediction did not come to pass, I was startled to learn that at the 2008 contest, the leading computer program came up shy of that mark by just a single vote. 2009’s contest could be the pivotal one. I decided to get involved, as one of the human “confederates” who talk with the judges and try to convince them that we are, in fact, human.

The computer program that does the best job each year of persuading the judges it’s human wins what’s called the Most Human Computer award, which includes the official bronze Loebner Prize medal and a small research grant for its programmers. But there’s also another award, bizarre and intriguing, for the human that does the best job of persuading the panel: the Most Human Human award.

One of the first winners, in 1994, was Wired columnist Charles Platt. How’d he do it? By “being moody, irritable, and obnoxious,” he says—which strikes me as not only hilarious and bleak but also, in some deeper sense, a call to arms:

How, in fact, do we be the most human humans we can be—not only under the constraints of the test, but in life?

2. Many people feel uneasy about the idea of computers passing themselves off as human beings. Do you?

One of the strange and fascinating things about the Turing test is that it cuts both ways: Alan Turing proposed his test as a way to measure the progress of technology, but it just as easily presents us a way to measure our own. Oxford philosopher John Lucas says, for instance, that if we fail to prevent the machines from passing the Turing test, it will be “not because machines are so intelligent, but because humans, many of them at least, are so wooden.”

Thus I see in the Turing test a way not only to think deeply about artificial intelligence, but an opportunity to think deeply about ourselves. Many people see the story of AI as a dehumanizing narrative: I see it as much the reverse. Most of the conversational systems that compete in the Loebner Prize are constructed to perform various types of simplifications of human dialogue. A look at the history of the test and at how these programs are built provides us, perhaps counter-intuitively, with a more detailed and nuanced appreciation for what it is about natural communication that’s so rich and complex. And it also gives us—in a time when we’re so often sacrificing that richness to laggy cell-phone calls and compressed text messages—some pointers for how to communicate and interact better.

3. What would you say is the most important thing you have learnt from your conversations with artificial intelligence software?

It’s hard to single out a particular thing, in that the whole process of researching the history of the Turing test and preparing for my role as a confederate was an incredibly enlightening journey. I not only looked at some of the great successes and failures of Turing tests past, but spoke with linguists, computer scientists, psychologists and philosophers about just what makes humans, and human communication, so distinctive.

I suppose part of the big takeaway for me is that ultimately I found the line between humans and machines to be blurrier than one might expect: partially because we have consciously modeled software in many cases on humans’ own thought process, and partially because we’re sometimes more “mechanical” or “algorithmic” in the ways we talk to each other than we might realize. The book chronicles the call center operators of today, and the switchboard operators of several decades ago, complaining that by being forced to respond to callers in rigid ways and using pre-formulated language scripted by someone else, they’ve found themselves in a position where they’re effectively chatbot programs— except the software is running on their own brains rather than on a computer. Similarly, another chapter describes what looking deeply into chess supercomputers like Deep Blue can tell us about what constitutes a good first date or a good reunion with an old friend.

So in some sense the crucial distinction between humans and AI isn’t about computers’ silicon vs. our carbon, or their microprocessors vs. our cortexes, but rather a question of approach: rigid, oblivious, streamlined procedure vs. an open-ended, plastic, flexible, attuned and sensitive state of figuring things out. We know intuitively that we can sometimes act “mechanically,” especially in conversation and sometimes despite our best intentions—the Turing test gives us both a concrete way to measure that, and a roadmap to avoiding it.

4. And the most surprising...?

In investigating the ways in which our human sense of self has been entwined with the history of computers, one of the most surprising discoveries for me was that computers used to be human. That is to say, in the early 20th century, before a “computer” was one of the digital processing devices that permeate our 21st-century lives, it was something else: a job description. Computers were (human) numerical analysts on the payrolls of universities and corporations; teams of computers were behind everything from the first accurate predictions for the return of Halley’s Comet to the Manhattan Project at Los Alamos. When Turing was first introducing the concept of the digital computer, it was by way of metaphor: “The idea behind digital computers may be explained by saying that these machines are intended to carry out any operations which could be done by a human computer.” Of course, now we know that the literal and the figurative have switched places, and in the 21st century it is the human math whiz that is “like a computer.’ We imitate our old imitators, one of the strange reversals of fortune in the saga of humans and machines.

Of course, another of the great surprises of the book was about surprise itself: namely, learning that the heart of information science—and arguably, therefore of communication—is surprise, and its mathematically-precise sibling, “surprisal.” Indeed, the concept of “information entropy” puts surprisal at the center of understanding everything from contemporary poetics to Oprah's interview technique, to how best to "catch up" with friends, and why novels are often regarded to contain “more” than their respective films, despite the file size of the book being orders of magnitude smaller.

5. There are so many strange and wonderful stories in your book. What have you enjoyed most about writing it?

Part of what I enjoyed so much about writing The Most Human Human was that, as someone who has backgrounds in computer science, philosophy, and poetry, I found myself in a rare position to be able to draw from and connect all three in preparing for my role as a human confederate.

More than that, it’s been a particular joy to work on this book as I found that not only was I exploring and chronicling a particularly intriguing thread in the history of science, but that what I was discovering in doing so had so many unexpected ramifications across daily life. My study of chatbots—these conversational imitators and mimics—and my conversations with linguists, poets, philosophers, and the like, provided me not only the set of strategies that I took with me to the Loebner Prize competition in Brighton, but with a lens for appreciating and savoring the unsung complexities of the interactions we have every day.