Future Of Work

Rival Robots & Unsupervised Algorithms

 

Last night I was listening to Martin Ford , author of the book, Rise of the Robots, Technology and The Threat of Jobless Future. The first writer to look deeply into the issue of technology-led unemployment, Ford’s vision for the world of work is both convincing and worrying.

 

It is old news that as the price of robots goes down and human capital goes up, jobs are being replaced by machines. What is alarming is the rate at which people are being replaced. A 2015 report by Ball State University shows that while manufacturing grew by 17.6% between 2006 and 2013, 88% this growth was due to advancements in automation and technology, increasing production. The growth did not correlate to new jobs and “employment during this period was largely stagnated.”

 

Those hoping that that manufacturing will return to America may see their wish as labor costs increase in nations like China, making it economically feasible for US companies to return to American soil. However, while industry may repatriate, don’t expect the number of workers required to rise dramatically. A factory that once may have hired thousands of workers will now hire hundreds. Robots take care of the rest.

 

The Boston Consulting Group project that “growth in the global installed base of advanced robotics will accelerate…to around 10% annually during the next decade.” In some industries, they expect tasks performed by robots to account for more than 40%of manufacturing.

 

The same threat exists for knowledge workers. Blockchain threatens tens of thousands of jobs in clerical work alone. Even reporting on such changes may one day be the work of robots. Already, machines report the results of sports events and give financial summaries. You can’t always tell a robot from a human author. It is only a matter of time before they add the analysis.

 

Right now, writers and analysts are able to topple the non-human competition, if only because algorithms aren’t quite as engaging. And while they work at speed, machines have high error rates. Mistakes can be discarded but they must first be ascertained–and that has proven to be costly. Robots are cheap, but their errors can be enormous and speedy.

 

The fate of the company, Solid Gold Bomb, provides a cautionary tale. A few years ago the company celebrated tremendous growth as part of Amazon’s marketplace. Solid Gold Bomb used algorithms to produce thousands of t-shirts with the “Keep Calm” slogan adopted from a propaganda poster of the Second World War era.  They imported 700 verbs to a computer programme and expected variations of the Keep Calm slogan that would appeal a broader range of customers.

 

Unfortunately,violent slogans were produced through an error in the company’s digital scripting process. Next to Keep Calm and Teach On, you might find Keep Calm and Rape On. Many of the shirts were immediately removed from their website and that of Amazon’s. However, because there was a “deletion queue” offensive t-shirts remained for sale despite the viral uproar that opposed them. The public outrage against this company and anyone who sold their offensive t-shirts had long-lasting impact. The company filed for bankruptcy in 2013.

 

Whereas in the past the use of technology has been limited to the vision of individuals who have mastered the complicated tools and languages of data science, machine learning means that the machines themselves become increasingly automatous, self-regulating and flexible. In a sense, we now have a second brain: the brain of the machine. While algorithms create efficiency gains, they also amplify error. Machines may be efficient, but they can also be efficiently harmful.

 

by Marti Leimbach

 

 

 

Previous Story
Next Story

No Comments

Leave a Reply