The Most Human Human: What Artificial Intelligence Teaches Us About Being Alive

By Brian Christian
Viking, 2011
320 pages (hb)
ISBN: 978–0670920808

Reviewed by Matt James

Given the growing interest in all things neuroscience, human behaviour, consciousness and advances in AI, I would be lying if I said I was not doubtful that this book could just be another looking at the relationship between humans and computers. Admittedly, from the strapline – “What Artificial Intelligence Teaches Us About Being Alive” this particular book did seem to be more intriguing than others. Nevertheless I was preparing myself for what might become a turgid read written by a computer or AI geek. I was pleasantly surprised with what I found.

The Most Human Human is perhaps the most engaging, fun, witty and provocative read that I have come across for some time.  The author excels at his craft and writes in a clear and lucid manner, taking complex ideas and thinking and making them easily accessible and understandable for the reader to comprehend. The effective use of illustrations and everyday examples aids this process. Christian excels at communicating detail and information as well as allowing the reader to learn at the same time. Reading the book is not a lesson in absorbing facts and details with little comprehension of how it all fits together. I wonder whether the author’s eclectic mix of literature and science training has something to do with this because this is not something that is easily found in other books of its kind. In the same way that you know the skilled keynote speaker who makes their talk so captivating and engaging has spent many more hours honing and refining it in private to make it that good, as you progress through the book you cannot help but think there has been much honing and drafting and skill deployed in producing a book of such lucidity. Hats off to Brian Christian for a book I have no hesitation in recommending!

Structure

The book consists of eleven chapters and centres around the author’s journey of discovery, exploration and preparation as he competes head–to–head with the world’s leading AI programmes at the annual Turing Test competition; the test devised by Alan Turing and widely accepted as the test threshold at which a machine can be said to be ‘thinking’ or ‘intelligent’. More than just a narrative of events of what he learnt along the way, Christian discusses key ideas which he identifies during the course of his preparation and takes time to comment and reflect on the implication of these discoveries in the light of advances in AI and what it means to be human.  Without doubt advances in the area of AI have been significant and one cannot live for very long in Western society without being more and more aware of ‘intelligent’ systems around them. Whether it is as you text on your iPhone using predictive text or engage with the online customer service supervisor with a query concerning your ticket booking online, AI is all around us it would seem.  But alongside this we must surely take time to consider what all this says about our humanity. Are we really no better at being human than the machines we’ve created?  If the likes of Ray Kurzweil are to be believed then once machines have surpassed us, the role of homosapiens will have become redundant as we merge with our machines.  There is an added poignancy to this statement when you consider that to date computers have come within a single vote of passing the Turing Test.

It is within this context that Christian takes time to reflect on what all this means to humanity. Rather than seeing AI as the demise of humanity, Christian’s premise is that it actually affords us  profound and exciting opportunities to rethink what it means to be human and what it means to be intelligent. He adeptly explores and demonstrates the advances in AI systems, acknowledging at the same time their weaknesses and offering a realistic assessment of the reach of their achievements into the future. In so doing, he allows the reader to wonder again at our most human abilities to learn, to communicate, to intuit and to understand.

Multidisciplinary

If this was not engaging enough, Christian helps to capture the reader’s imagination by drawing upon illustration and ideas from a broad spectrum of disciplines including science, philosophy, sociology and the arts, making this an excellent book to span the disciplines and perhaps stimulate further multidisciplinary engagement with the issues the book raises.  At one point in the book you find Christian discussing Aristotle’s Nicomachean Ethics, only to then find him exploring the intricacies of language and language entropy later on!

A real life dimension is also brought through illustrations from work, chess, speed–dating, psychiatry and video games. For me, the chapters exploring the world of and information entropy were particularly engaging.  In chapter 5, Christian focuses particularly on chess and the victory of Deep Blue computer over grand master Garry Kasparov in 1997. He helps the reader to realise and appreciate the contrasts between human and computer by exploring where the computer’s greatest strength lies in a game of Chess. The computer is strongest in the opening and closing sequences of the game and where humans rely most heavily on memory. By opening the reader (or at least non–Chess playing reader) to a greater understanding of what is known as “the book”, which constitutes the knowledge of these tried and tested moves, Christian proceeds to demonstrate the unique function humans can bring to the game. By the middle of the game, the sheer number of permutations of moves is just too much to memorise. Therefore, chess players need to abandon the moves – get “out of book” – and begin to act unexpectedly, something which computers can struggle with. As is the style and character of the book, the author demonstrates this idea of acting “out of book” with everyday life examples including letter writing and human interactions.

Information entropy

Chapter 10 – High Surprisal  – explores the creative imagination and intuitiveness of humans.  Christian explores information entropy and the implications of the Shannon Game. Not to be confused with thermodynamic entropy, entropy in this context refers to the theory that the higher the entropy, the more information there is to obtain.  The Shannon Game model seeks to guess the next letter of a text and the logarithm of the total number of guesses required to give you entropy of that passage.  The idea is to estimate how much knowledge native speakers bring to a text. Using this example Christian discusses in a fascinating way the workings of predictive text and AI ‘intelligence’ concluding that computers do not respond to your sentences so as much as they simply complete them.  Perhaps the following quote helps to offer a taste of Christian’s perspective. “Cobbled–together bits of human interaction do not a human relationship make…..Not 50 one–night stands, not 50 speed dates, not 50 transfers through the bureaucratic pachinko…..Fragmentary humanity isn’t humanity.”

Conclusion

What this book does which other similar publications may fail to do is to provoke you to think about the questions it raises after you have finished reading the book.  Every time I go to speak to someone who I have never met or who I have known for years.  When I go to Google and begin entering the first three letters of the search term I want to use. I cannot forget what I have read in The Most Human Human and think further on the implications of my human interaction with AI and my interaction with other humans.  

Far from assuming a position which laments the advent of AI and computers and ‘Humanity is doomed’ state of affairs, I believe Christian’s book helps to inspire and motivate on all fronts. Inspire and motivate further advances in the area of AI systems and in so doing inspire and motivate us as humans to consider afresh what it means to be human and to explore to a much fuller and richer dimension the extent of our humanness.

• AI & Robotics

There have been 0 replies to this Article. + Post your comment here.