Skip to main content

Artificial Communication - Elena Esposito ****

Sometimes, you read a book where the content is so interesting and important that it's necessary to disregard the fact that it's not a great piece of writing - this is such a book. The fundamental point being made here is that the concept of 'artificial intelligence' is mislabelled. What a smart speaker or AlphaGo or an autonomous vehicle does has nothing to do with intelligence. Intelligence implies understanding - but the algorithms of AI have no understanding of the data they handle. Instead, sociology professor Elena Esposito suggests, what they are capable of is artificial communication.

We see this most clearly with those smart speakers, or voice assistants on smartphones. I ask Siri a question and she responds - it is a form of conversation, and one that mimics intelligence, but it is based on something totally different from understanding. The same can be said when using a search engine. These are no longer mere indexes - we ask them questions and they reply with answers. But even in a less obvious application, there is a form of conversation going on between the AI and its environment - enabling it to carry out actions that in a human require intelligence, but don't in the way that the software/hardware is implemented.

For me, this is absolutely fascinating - a necessary adjunct to reading about AI to understand its limitations and how to deal with those limitations better. That's why the book gets four stars. Unfortunately, though, it would have been a lot better if Esposito had taken on a co-author. The text is highly repetitive, making some points repeatedly, and that's in a book that's already short at 111 pages before hitting the notes.  It is also frustratingly vague. Many of the aspects of AI it talks about are covered in a broad, high level fashion without clear examples. The writing sometimes has the feeling of the kind of soft (indeed, squidgy) social science paper that was the target of the Sokal hoax.

The worst example is a chapter on data visualisation and visual text analysis which does not contain a single specific example. The only illustration in the whole chapter (indeed, pretty much the only one in the whole book) is a picture of a house to illustrate the high bandwidth of a picture. But in a chapter crying out for some actual visualisations (and more of an explanation of what digital text analysis actually does with clear examples) there is no assistance for the reader.

Frustrating in some ways, then, but a book I still heartily recommend because the topic itself is so illuminating, even if the approach taken sadly isn't. We can't expect every academic to be a great writer - many aren't - but when that's the case, a much better book could have been produced if they had paired up with a professional.

Hardback:   
Kindle 
Using these links earns us commission at no cost to you
Review by Brian Clegg - See all of Brian's online articles or subscribe to a weekly digest for free here

Comments

Popular posts from this blog

David Spiegelhalter Five Way interview

Professor Sir David Spiegelhalter FRS OBE is Emeritus Professor of Statistics in the Centre for Mathematical Sciences at the University of Cambridge. He was previously Chair of the Winton Centre for Risk and Evidence Communication and has presented the BBC4 documentaries Tails you Win: the Science of Chance, the award-winning Climate Change by Numbers. His bestselling book, The Art of Statistics , was published in March 2019. He was knighted in 2014 for services to medical statistics, was President of the Royal Statistical Society (2017-2018), and became a Non-Executive Director of the UK Statistics Authority in 2020. His latest book is The Art of Uncertainty . Why probability? because I have been fascinated by the idea of probability, and what it might be, for over 50 years. Why is the ‘P’ word missing from the title? That's a good question.  Partly so as not to make it sound like a technical book, but also because I did not want to give the impression that it was yet another book

The Genetic Book of the Dead: Richard Dawkins ****

When someone came up with the title for this book they were probably thinking deep cultural echoes - I suspect I'm not the only Robert Rankin fan in whom it raised a smile instead, thinking of The Suburban Book of the Dead . That aside, this is a glossy and engaging book showing how physical makeup (phenotype), behaviour and more tell us about the past, with the messenger being (inevitably, this being Richard Dawkins) the genes. Worthy of comment straight away are the illustrations - this is one of the best illustrated science books I've ever come across. Generally illustrations are either an afterthought, or the book is heavily illustrated and the text is really just an accompaniment to the pictures. Here the full colour images tie in directly to the text. They are not asides, but are 'read' with the text by placing them strategically so the picture is directly with the text that refers to it. Many are photographs, though some are effective paintings by Jana Lenzová. T

Everything is Predictable - Tom Chivers *****

There's a stereotype of computer users: Mac users are creative and cool, while PC users are businesslike and unimaginative. Less well-known is that the world of statistics has an equivalent division. Bayesians are the Mac users of the stats world, where frequentists are the PC people. This book sets out to show why Bayesians are not just cool, but also mostly right. Tom Chivers does an excellent job of giving us some historical background, then dives into two key aspects of the use of statistics. These are in science, where the standard approach is frequentist and Bayes only creeps into a few specific applications, such as the accuracy of medical tests, and in decision theory where Bayes is dominant. If this all sounds very dry and unexciting, it's quite the reverse. I admit, I love probability and statistics, and I am something of a closet Bayesian*), but Chivers' light and entertaining style means that what could have been the mathematical equivalent of debating angels on