"Will we be retired — or unemployed?", speakers sometimes ask at technology conferences that, such as the Singularity Summit, contemplate a future in which artificial intelligences are a lot smarter than we are. In Christopher Steiner's Automate This: How Algorithms Came to Rule Our World, it becomes clear that the intelligence doesn't have to be superhuman: the rather more mechanical algorithms and programs we have today are running things in industries as diverse as finance, medicine, and entertainment.
Humans still make the final decisions, but many of the mid-level people — the technicians who read scans, the A&R folks who pick out new artists to invest in — may be on the way out. Algorithms, argues Steiner, do a better job of predicting chart-topping songs, good investments, rare illnesses and political crises than even the most experienced and highly-trained humans.
Steiner opens with the big data equivalent of a joke: the 2011 incident in which the sale price of a $40 used book was bid up on Amazon by competing bots to more than $23 million. Then comes less of a joke: the 2010 'flash crash', in which automated trading briefly bonked the Dow average on the head. And then he plunges into history with the little-known figure of Thomas Peterffy, who first figured out that a properly primed computer could trade a lot faster than his human competitors. That discovery and its aftermath launched what Steiner calls "the 30-year tale of creeping algorithmic takeovers".
It's clear that Steiner expects the spread of algorithms to continue throughout the rest of human endeavours. This is less clear to me. For one thing, in many of the cases he covers there hasn't been much time to observe what happens next, or what kinds of failures we may find in the longer term. He does note that genuinely new and inventive music or movies may be passed over by algorithms that derive their ability to spot the gems in new work from findings identifying the basic characteristics of past hits. Still, humans can keep adapting systems in response to such failures. It didn't, for example, take many months after the flash crash for the markets to install 'circuit breakers' that halt trading if there are sudden large rises or falls indicative of runaway automated trading.
Where humans may get lucky lies in the problems that engineers think are worth solving. "The best minds of my generations [sic] are thinking about how to make people click on ads. That sucks," Harvard-educated mathematician Jeffrey Hammerbacher, tells Steiner. Hammerbacher's own career led from crunching data on Wall Street and Facebook to the chief scientist job at Cloudera to create data analytics software that can be applied to any problem, earth-shaking or pointless.
This is the hopeful note on which Steiner concludes: alienated from Wall Street by the 2008 crash and eager to do meaningful work with an impact on the world, many more of today's generation of young engineers will be moved to tackle important problems rather than solely lucrative ones. Maybe one of those problems will be finding employment for all those displaced folk.
Automate This: How Algorithms Came to Rule Our World
By Christopher Steiner