At the end of the 19th century Americans and Europeans started eating horses. As a result of technological innovation many horses had lost their jobs, oversupply pushed horse meat price down and market forces drove our four legged friends from the stable to the table. The New York Times noted in 1895 that 'the time of the horse’s great usefulness is past and the only solution will be found in introducing the meat as an edible.'
Actually, the steam engine had created at first a huge number of job opportunities for horses, as cities grew larger and ports and railroads construction called for traction power and the ability to move on irregular terrain. So the number of horses in the US grew from 5 million in 1840 to 22 million in 1900. However, steam engines soon became efficient enough to automate the horse job as a long distance mover of people and goods. Then came electricity and hundreds of thousands of city horses employed in urban transportation got fired. And finally, the internal combustion engine established the new standard for private and commercial transit.
It took time for technology innovations to spread across the economic system and society, but eventually the impact on the horse population turned out to be dramatic: from 22 million in 1900, there were 4 million horses left in the US by 1960. A clear case of technological unemployment, where the market progressively replaced the horse with more productive machines and found the poor animal an alternative… occupation.
Over the last 30 years, a new wave of radical technology innovation has been unfolding. Today the applications of advanced robotics, artificial intelligence, sensing and infinite computing such as self-driving vehicles or big data are providing a much more productive alternative to human work in the completion of all sort of professional tasks.
Everybody agrees that an increasing number of our jobs is being partly or entirely automated out of existence. Most projections suggest that between 40% to 50% of all existing jobs will be impacted over the next 20 years. All around us, assembly line workers are being replaced by the likes of Baxter. Supermarket cashiers, bank tellers, travel agents, call center operators, accountants, farmers, and secretaries are turning into algorithms. Machines are outperforming doctors, legal practitioners, and stock traders. Computers are even getting pretty good at writing articles or music (granted, not great stuff… but most people cannot tell the difference).
There are various opinions on whether technology will generate enough new jobs to compensate for those it automates, as it happened in the past when manufacturing replaced agriculture as the main global employer. In fact, this time, things may be quite different: the pace of innovation and its impact on the economy is faster than ever, technology is surgically targeting the replacement of human beings and the new jobs are a fraction of those destroyed and typically higher skill/higher pay. So the result is that the labor market doesn’t adjust, shrinks and gets increasingly polarized.
Major economic and social challenges are emerging. When machines do most of the work and salaries are no longer paid to people, who is going to buy products and keep the system going? How will we deal with the individual and social distress related to widespread mass unemployment and its political consequences? In practice, when 20 million of driving jobs in the US and Europe are replaced by a few thousand programmers and data scientists, what are we going to do with taxi drivers? Are we going to eat them?
The answer is that governments should seriously start debating, designing and experimenting policies to maximize the social benefits of digital innovations while managing the impact of the transition. Let’s rule out the idea of slowing down innovation to protect existing jobs. Companies, industries or countries taking this route would soon lose competitiveness in the global markets and consensus for not responding to customers and constituents increasingly sophisticated expectations. This leaves us with fundamentally 4 policy areas:
- Sharing paid work may be a short term option worth considering. Increasing vacation time, bringing forward retirement or shortening the working week could cushion the initial impact of automation. Shorter weeks, for example, are used by some companies to attract rare talents while early retirement may provide opportunities for valuable social work.
- Regulation to create new job opportunities is also a potentially useful policy tool. Traditional tax breaks and subsidies for companies that hire new employees, however, are unlikely to work: with an industrial robot 4 times more efficient than a skilled worker it would be difficult for the government to match the economic benefits of automation. On the contrary, providing strong support to entrepreneurship by relaxing red tape, financing business incubation and venture capital, backing access to debt or helping small companies go international are all policies bound to generate structural economic and social benefits.
- Education has a big role to play both in the short and long term, provided policy action is taken to revisit its organization and content. Education needs to be made available when needed in life to catch up with a rapidly changing workplace and delivered across the most efficient channels. Good ideas for policymakers include, for example, MOOCs and mini colleges. But perhaps more importantly, the content of education needs to change. Higher education should increasingly focus on the critical role of science, technology, engineering, and mathematics (STEM) in preparing the workforce of the future. Children education should adapt to an information-saturated world teaching how to handle information, or in other words how to search, read, critically analyze and believe. This implies reconsidering profoundly the role of teachers, classrooms, and traditional subjects.
- Finally, a clear long-term policy answer to structural unemployment is providing everybody with a minimum income. If salaries are no longer paid and money is not fed back into the system demand dries up and the game is over. So it makes a lot of sense arguing that the same companies that automate jobs should fiscally contribute to the financing of a minimum income policy. To ensure that the minimum income does not create frustration and marginalization, however, it needs to be enriched with a system of incentives to promote active participation in economic and social life.
At the end of the day, those policies may well be preparing individuals and society for a very different future, where salary is no longer the main incentive for being an active and productive member of your community nor the number one measure of success and recognition. Take horses. In the second half of the 20th century their number started increasing again and today there are over 9 million heads in the US, up from 4 million in 1960. Horses no longer need to work hard transporting goods and people or fighting wars. Rather they provide a socially valuable contribution to human well-being as sport partners, therapy animals, or simply companions of a healthy lifestyle. Perhaps digital innovations will modify our economic relationships too and eventually re-balance our values and beliefs.