Book Review: The Second Machine Age

In The Second Machine Age Erik Brynjolfsson and Andrew McAfee eloquently document many of the changes that increased Artificial Intelligence is bringing. They go on to make a series of policy recommendations that, whether you agree with them or not, will educate you further about the economic and social changes that increased Artificial Intelligence will bring.

Artificial Intelligence - the present and future

The Second Machine Age begins by documenting many of the changes that are happening to economics and society as a result of increased automation. Through examples of computers beating humans at Jeopardy!, unbeatable chess computers and cars that can drive themselves Brynjolfsson and McAfee demonstrate that the world is changing and will change radically over the years to come. The book is ambitious in that it initially seeks to document the impact of automation mostly through third-party research, theorise about the future impact of Artificial Intelligence and then make policy recommendations for legislators.

The book draws on a large body of academic research and cites several studies around the impact of Artificial Intelligence. It is not exploring what is possible, but rather showing (and sometimes in wonder) what AI is capable of. This includes boyish excitement of being driven around in a car with no driver or celebrating that Watson can beat humans at Jeopardy! every time. There is some reflection and analysis given to the social and economic impact of AI during these examples but predemoninantly the initial chapters of the book can be summarised by “this is what’s happening, it is amazing, and we’ll need to work out how to deal with it”.

Productivity Challenged

Historically GDP has measured a country’s productivity and to a large extent continues to do. GDP is inherently labour based and the idea that a nation’s standard of living equates to GDP output still governs much of how the world is run. The book points out that because of digitisation and automation this is becoming increasingly less relevant and it is no longer the best way to measure growth.

Production in the second machine age depends less on physical equipment and structures and more on the four categories of intangible assets: intellectual property, organizational capital, user-generated content and human capital.

The authors suggest using consumer surplus to determine growth because it offers a better representation of improvements in quality of life as well as monetary growth. Many new technologies are free to users, such as Google, Wikipedia, and many digital technologies which drastically improve quality of life but are not accurately reflected in GDP numbers. This was a really interesting section of the book and effectively challenges many of the ways that we measure productivity and ultimately human happiness.

Reflection on Wealth

The authors reflect that far from the vision of a distributed, egalitarian platform the web has increased the gap between rich and poor. The cause of this is suggested to be a result of both digitisation and globalisation. The book notes that the top 1% of earners in the USA have seen large increases to their income whilst the lowest earners have seen their earnings decrease. The authors cite the example of digital music where more music is being listened to more than ever but artists have seeen a dramatic drop in revenue income. While it now costs almost nothing for distributors to sell digital music the music industry has had to accept that the new distributors like Spotify and Google Music will take a greater cut of the overall revenue pie and that record labels and artists will get less.

The authors describe this as bounty and spread, where distributors or technology providers command a bounty for having understood how to solve a particular problem but relatively everyone else’s economic position in the problem will decrease.

Human Jobs will be automated

The book points out the inevitable truth that advances in AI will result in job losses. The example given in the book is of professionals that prepare tax returns. Software is already largely automating this task and it is cheaper and more efficient for the buyers of this service to use software services.

If you can give precise instructions to someone else on exactly what needs to be done, you can often write a precise computer program to do the same task. In other words offshoring is often only a way station on the road to automation.

In a largely utopian view the book’s authors urge legislators to let AI and net effects like this to happen. The underlying message of the book encouraging humanity to run with machine rather than to fear it.

The book seems to position itself in a space that welcomes the advances that AI can bring to humanity but is acutely aware of the need to create regulatory and legislative environments for it to flourish. Indeed there are two entire chapters devoted to Policy Recommendations and Long-Term Recommendations. These recommendations support the underlying conclusion of the book that humans should focus on skills that compliment machines. They should not seek to fight automation and moreover the authors are hugely optimistic about what increased AI can bring. The most telling example in the book is of human chess players being able to beat a chess supercomputer with the support of an average chess computer. The authors see that combining the human consciousness with Artificial Intelligence can propel intelligence beyond what can be achieved without it. Far from running from machines the book recommends we should embrace Artificial Intelligence and recognise it as a compliment to our own human abilities.

Whether you agree with the liberal message to encouraging Artificial Intelligence and letting it run its course the book raises many issues that Technologists, Academics and Legislators should be considering. This book is not the only book to do this but if you are looking for a book to challenge your social and economic assumptions on Artificial Intelligence the book achieves well.


Can you help make this article better? You can edit it here and send me a pull request.