New new things

We are enthusiastic monkeys. When we make a new discovery, we rush to see how we can use it in every domain of our lives.

Can we eat it? Can we stick it into things? Can we hit other monkeys with it?

When electricity emerged as a new technology, it gave rise to electric baths, electric corsets and electric hairbrushes.  There was even the electric Temple of Heaven & Hymen Celestial Bed[i].

The discovery of Radium resulted in its use for around 150 medical complaints, as well as use in cosmetics, tonics and toothpaste[ii].

We are going through the stick it into everything phase now with AI and it’s producing some extravagant claims. Our monkey enthusiasm is also being stoked by people who expect to make vast amounts of money from this new marvel.

There are many discoveries and inventions that were ahead of their time or at least beyond the other capabilities of their time to properly exploit – the first solar cells were invented in the 1880s and the first electric car even earlier[iii]. An invention can be good, or even great, but may still take a long time to land.

There are more prosaic examples of good ideas that failed or floundered because not all of the necessary ingredients for success were yet in place. The paperless office is only now becoming a reality, having been feverishly discussed since the late 1990s and even electricity took nearly half a century to move from initial commercial exploitation to near ubiquity[iv].

In our conceit that we are so advanced, and they were so primitive, we mock the earlier failures without seeing that we may be repeating them.

Not the Messiah

The idea of creating a new form of intelligence is an awe-inspiring one. When this started to look like a real prospect, it was called Artificial Intelligence (AI) and more recently it has been called Artificial General Intelligence (AGI), but over time what is meant by these terms has been subject to a form of shrinkflation, where what they end up describing is considerably smaller than what was originally envisaged[v].

Which is not to say that what we now call AI is unimportant or that it will not have significant consequences at personal, commercial and societal levels.

Over the course of this book, I will try to tease these out and separate the true and the possible from the nonsense and the propaganda.

This is a book by an AI Expert 

I believe in value of expertise but when people talk about AI experts, this group seems to include many who think they qualify because they wear shorts and flip-flops to the office, and work in front of a screen[vi].  

It also includes lots of people selling things. 

I’ve read many books by people who hold themselves up as AI Experts and have spoken to quite a few of them. Some of these people are well qualified in terms of years of study or relevant work. Many are not.

It’s also worth saying that the answers to many of the most important issues around AI – What can I use it for? How might it develop? Is it going to kill me? – have answers that relate less to the mechanics of how the technology works and more to understanding questions of logic, economics, sociology and even philosophy and language. 

There are many clever people, and even some wise ones, working in the very broad field of what, by one definition or another, is described as artificial intelligence, but these are not always the people being heard.

In writing this book, I am trying to draw out those voices and also apply useful insights from other fields.

So, when I gingerly hold myself out as an AI expert, it is in part from recognising myself as being on the flip-flop end of the spectrum through running an AI start-up, but it is mostly as someone who is communicating the vast and deep AI expertise of others.

More usefully

I’ve also spent many years running various risk management consultancies, mostly specialising in board level risks. Basically, the big company-killing things. My technical competencies cover risk management, governance, law, accountancy and other equally kick-ass, rock and roll areas. 

The more practical skills I’ve developed come from being paid to spend time listening to what people say is going on, and then working out what is actually going on from what is said and unsaid, from what I can find out elsewhere and from my understanding of motives, organisational dynamics and market mechanics[i].

In the field of AI, I often feel a dissonance between what I am told and what I can see.

I believe that artificial intelligence is important and that it has potential to bring great and useful changes, but it is surrounded by a great cloud of nonsense.

This book is my attempt to cut through that fog.

Read this quickly

Some of the detail in this book will inevitably date quite quickly in the fast-moving environment of AI development but my intention throughout is to discuss ideas that have described, and will continue to describe, the broad principles and context of what is going on.

More generally:

The idea of this book is to explain what AI actually is, and to do it in a way that is comprehensive but still reasonably concise.  

The intention is to give you[ii] something that clarifies what’s going on and perhaps even helps you to get the best out of the technology.

I’ve also tried to set AI in its wider context[iii] and to draw on insights from historical and current thinkers.  Some from the AI world but many are from other fields.

About the footnotes

This is a very ambitious undertaking, and I have tried to cover a lot of ground in a way that is crisp and entertaining. This has meant truncating some of the arguments.  

Where I have done this, I’ve tried to provide more detail and links to my sources in the footnotes, So you can dig deeper if you want. 

I’ve also used the footnotes as a place to quarantine my more caustic asides, bad jokes and obscure references, and all the interesting digressions that I thought would interfere the flow of the main text

Footnotes to the Introduction

[i] I also like the observation made by Gary Marcus, a psychologist and cognitive scientist, and a prolific commentator on AI, who describes his own standing as being like Joseph Conrad, who had insights into English because he was a non-native speaker.  

Quoted in Architects of Intelligence; a book of interviews conducted by Nick Bostrom.

[ii] Yes, you dear reader.

[iii] Society, business, work and, dammit, life.

[i] 7 Shocking Uses for Electricity – The Saturday Evening Post 21st October 2021

[ii]The shards of heaven beneath our feet – The Spectator – Guy Stagg – 1st of February 2025. A review of Under a Metal Sky: A Journey Through Rocks by Philip Marsden.

[iii]8 Inventions That Were Way Ahead Of Their Time – Business Insider - Dylan Love, 3rd of July  2013

[iv]Why I think AI take-off is relatively slow - Marginal Revolution – Tyler Cowen, 23rd February 2025. The paperless office observation was from Ricardo in the comments.

[v] Much, much more on this later.

[vi] This is particularly common in polls of AI Experts. When you start digging into the methodology, it is apparent that many just happen to work in various parts of the tech industry and have limited claims to expertise around the topics being surveyed. Occasionally it seems about as meaningful as asking Homer Simpson about the safety of nuclear power plants.