I’ve set myself the goal of writing a short book review every week in 2020. Let’s see how that goes. Happy New Year!
Daniel Susskind’s A World Without Work is, primarily, an excellent guide to the economics of automation and to the latest progress in artificial intelligence. Susskind begins by describing “a history of misplaced anxiety” about the machines taking the jobs, before outlining the influential Autor-Levy-Murnane (ALM) paradigm of 2003.
ALM emphasise tasks, rather than jobs: automation is far more likely to encroach on a narrow task (such as adding up the prices of goods at a supermarket checkout) than to completely replace a job such as a checkout assistant. We should therefore expect automation to reshape jobs, not replace them.
So far, so good – and Susskind’s contribution is to deliver a crystal-clear explanation of the received wisdom in economics, with plenty of examples. It’s a model of popular academic writing.
Susskind then moves to argue that many economists are underestimating what automation is now achieving. The ALM idea of “routine” and “non-routine” tasks is starting to break down – consider the progress in image recognition, legal document analysis (something Susskind has studied deeply), or translation. Is Google Translate really performing a “routine” task? What about AlphaZero, the self-trained system that destroyed the best Chess and Go players in the world, human or computer?
Susskind’s point is that the ALM paradigm needs rexamining: we can no longer simply assume that large numbers of tasks are “non-routine” and therefore robot-proof. Neither can we assume that almost all humans will find it straightforward to earn a living. We need to adapt to a world where technological unemployment may arrive on a large enough scale to cause real misery and disruption.
Finally Susskind reviews solutions, such as a basic income, education, and – speculatively – a “meaning-creating state”, by which he means a state that is able to produce a sense of purpose, meaning and identity that in the 20th century was provided by our careers. I think he’s right to identify the goal of helping people find a sense of meaning and identity; I’ve no idea, however, what a “meaning-creating state” would really look like. But perhaps that is less a criticism of Susskind and more a recognition of how deep and complex the challenge might become.
Recommended!
Compare Carl Benedikt Frey’s The Technology Trap, a book which is intimidatingly weighty but is well-written and accessible. Frey was one of the researchers behind the viral “xx% of jobs are vunerable to automation” claim, but this book is much more than a book about robots taking jobs – it’s a history of automation from pre-industrial times.
So far I’ve only read the (penultimate) chapter on artificial intelligence; it’s excellently written, full of examples and studies I hadn’t previously encountered, and I learned a lot. Not obviously contradictory to Susskind’s book, and it is intriguing that there are so many ideas out there that the overlap is modest.
Catch up on the first season of my podcast “Cautionary Tales” [Apple] [Spotify] [Stitcher]
My book “Fifty Things That Made the Modern Economy” (UK) / “Fifty Inventions That Shaped The Modern Economy” (US) is out now in paperback – feel free to order online or through your local bookshop.