Bill Gates, for example, has called for the taxing of robots that take away jobs. This has elicited responses from leading economists, such as Larry Summers (former Vice President of Development Economics, Chief Economist of the World Bank, and US Treasury Department official) who argue against the idea saying that robots are job creators and that the idea of taxing them is profoundly flawed.
The focus in these debates is misplaced. Jobs are not created or lost because of a single technology, but because of the business models designed to leverage the power of the technology. Uber, for example, may be called a “taxi-hailing app” service but the business is a constellation of applications including algorithms, the automobile and GPS – all of which are organised around a single business model.
We’ve seen a similar example in history, with recorded music in the last century. It wasn’t the 1930s recording technology itself that threatened the jobs of live musicians. It was its combination with radio broadcasting, jukeboxes and the way businesses operated that led to job losses. Hotels, restaurants and bars replaced live musicians with jukeboxes. The coin-operated machines were cheaper and did not involve dealing with the demands of unionised musicians. A single recording could be played over and over without requiring the appearance of musicians.
As I argue, in “Innovation and Its Enemies: Why People Resist New Technologies”, the early recording of music destroyed the jobs of some live musicians and undermined their claim to property rights. The social objections became largely about monopoly power and less about the technology itself.
The technology did make huge gains for the music industry because of its ability to access a wider section of society. Small bands and minority musicians who could not have access to large markets were able to use the technology to reach niche audiences. More importantly, the spread of the technology made it possible for new genres, such as bebop, of music to emerge and eventually enter mainstream markets.
However, while history helps us to learn from the past, it has become a poor guide for emerging trends. This is because of the qualitative differences between discrete technologies that defined the Industrial Revolution of the early 1800s and today’s machines and platforms.
These fundamental differences – between past automation activities and today’s artificial intelligence – suggest the emergence of new economies operating under different ruleswhose contours are still sketchy.
The impacts will be strongly felt, given the integrated nature of the global economy, the rapid rate of technological change and the uncertainty created by technological abundance – which makes it difficult to predict where new ideas might come from.
Job creation – or loss – has to be considered in the context of the overall business.
This is best illustrated by looking at the difference between recorded music then and robots now. I have identified four big ones.
First, robots are being adopted at a much faster rate than recorded music was. Competitive pressures in industry are forcing entrepreneurs to look into deploying technological systems that enable them to stay ahead of the curve. Chinese manufacturers, for example, are responding to wage increases with one of the fastest rates of industrial robot adoption in the world.
Second, the consequences of robots are likely to be felt across global value networks triggering large-scale technological anxieties as workers fear that their jobs will be taken. This is partly because many of the leading industries rely on supply chains that are located in different regions and countries. Boeing, for example, sources parts from various parts of the US, Europe and many other parts of the world. Changes in its manufacturing practices involve coordination across those regions. This also applies to less complex products such as consumer goods.
Third, robots are advancing exponentially while human learning occurs at a much slower linear pace. Their rate of learning doubles in a short period while human learning is incremental and slow. This is partly a result of technological abundance and the growing ability for machines to teach each other how to improve the functioning of their algorithms. 3D printing, for example, is a combination of pre-existing mechanical technologies which now benefits from advances in digital technologies. It can now be diversely applied – from engineering to medicine. The more new technologies that are created, the greater the prospects of creating new applications. In many cases innovators start off with searching and using what already exists in novel ways before they invest in new research.
Finally, the effects of automation are likely to be felt over very short periods, compounding public concerns and leaving little room for adaptation. This is mainly because machines are reaching a point where they are learning to perform new tasks faster than workers can be retrained.
What we need are inclusive social policies that take into account faster access to emerging technologies, greater support to new businesses and more open dialogue about how poverty and inequity amplify the negative effects of new technologies.
This requires a much deeper look at how social systems and technologies shape each other to create more just and resilient economies.