The Most Human Human

Brian Christian

Why I looked at this book

I read about The Most Human Human in Tim Harford's Messy. Now while I think we should be suspicious about the claims of AI, the message seemed to be that we should be constantly competing with each other to be the most 'novel'. I feel that maybe being formulaic is useful sometimes. So I thought I should read this book to see what Brian Christian, The Most Human Human himself, had to say on the matter.

First impressions

The first chapter describes the origins of the Turing test, and points out that it's trying to do what humanity has been trying to do for millennia, that is figure out what makes human's unique. Christian also tells of his approach to his participation in the Turing test. Rather than follow the advice 'Just be yourself', he spent the months beforehand preparing for the test. That ties in with my feeling that trying to be more human is really being over-competitive. But then the reason the book exists is that he won, so I don't want to be prone to sampling bias. I'll have to see whether this competitiveness is a feature of the rest of the book.

Main review

I felt that the complaints against modern life being too formulaic weren't really the main point of the book, or at least if they were then the arguments weren't very persuasive. For instance, there's the behaviour of pick up artists (PUAs), who use very formulaic methods to chat up women, as described in Neil Strauss's The Game. It seems that by abandoning more human style conversation, the PUAs gain an unfair advantage. But reading further it begins to look like the PUAs are simply playing a different game, and the more human style conversation was really just a different script.

Christian also tells of a friend who became skilled in a part time job as a barista, getting to know how to deal with the various demands of the job. When she started her full time career, however, it felt more like she was following a script for much of the time. Rather than see this as a sign of the times, I'd see it more as the result of professional status. You become qualified for a certain line of work, and are expected to produce a consistent output, rather than being challenged by every new task you undertake.

I would say that, in fact, the book would be more useful to someone trying to improve artificial intelligence than to those fighting against it. Christian points out many of the shortcomings of existing AI's,which would seem to me to provide material for those working on AIs. I felt that this discussion was the most interesting part of the book. For instance some chatbots can be very convincing by just responding to the last sentence someone has typed. The bots don't maintain an internal state based on what has been said so far. Now in the first half of the 20th Century Behaviourism, which said that humans operated on this stimulus-response basis, was a popular idea. My feeling is that the coming of computers, with a definite internal state, helped to reduce the appeal of behaviourism. So I find it intriguing that computers look more human when they abandon an internal state.

Towards the end of the book, Christian puts forward the idea that as computers get better at doing what humans do, humans should try to improve what they do so as to stay ahead, rather than thinking "well that's it". I think that this is a reasonable point of view provided it doesn't mean that life is one long competition. We're already constantly told that we need to be economically more competitive. As computers can do more, we should be able to use our time for better things, rather than feel that we have to compete with them. If using our time for better things is the intention of what Christian writes, then that is good, but sometimes it's hard to tell. I feel that although it is certainly an interesting read, the book (maybe inadvertently) pushes for a more competitive mindset that is really necessary.