Automation of the workforce

I just read The Lights in the Tunnel: Automation, Accelerating Technology and the Economy of the Future, an exploration of the possible impacts of automation on the economy. The author, Martin Ford, proposes that the accelerating automation of jobs will lead to wide-spread unemployment as computers and robots take over many of the tasks currently completed by humans. This loss of income would significantly reduce the size of the consumer market, which would in turn reduce demand for products and push the economy into a downward spiral.

While the idea is not new (see Luddites), Ford’s main point is that few economists have explored the possible implications in depth. Much of the book outlines the possible consequences of automation, which in my view are very difficult to quantify. A legitimate concern for governments around the world will be the impacts on their tax base as more and more income is generated by software companies, automated warehouses and other online products which rely on limited human intervention. Without human salaries, there is little to tax. Since automation can be located anywhere, companies will lean towards states and countries that offer the lowest tax rates.

To compensate for lost revenue, Martin Ford proposes a gross margin tax, basically a tax on operating profit. He also proposes a sliding scale for deductions of salaries. In his mind, a company would be able to deduct 200% of a 50k salary and not be able to deduct anything above a 400k salary. This would amount to a stunted salary cap and could be a very interesting solution to the growing income inequality in our society. I think both ideas should be investigated in greater depth, the book does not however do so.

While I enjoyed Martin Ford’s exploration of the potential impact of automation on the economy, I do think his timeline is too fast. Technologists and IT people often only see the tremendous automation in their fields and extend it to the entire world. Their lack exposure to some of the more manual jobs – construction, retail sales, product distribution – push them towards very fast timelines (and libertarianism, though that’s another story). The vast majority of the global economy is not software or robotic.

What automation will do is to free up labour for other tasks. Rest assured, there is no shortage of work to be done in this world. The real question is how to allocate resources and motivate companies and people to work in fields that might otherwise not be highly profitable. Lest we forget, as late as 1780, over 90% of society was directly employed in agriculture, today less than 2% of workers create far more food than we ever thought possible. Yet, unemployment is still relatively low.

The reality is that as automation increases we need to put in place mechanisms to train people for other jobs and allocate resources to employ them. It is said that there are two ways to tie up large male populations in developing nations – the military or construction.

The problem of automation and its effect on the tax-base are real. California is home to Apple, Google, Facebook and other tech giants but cannot balance its books, this might very well be the canary in the coal mine. As Bill Gates recently discussed at TED, we need to reconfigure government to ensure tighter oversight and better allocation of resources. Part of the solution is open-data and open-governement (that’s another story).

While I do believe that we need to seriously reevaluate our taxation system and the way government works, I would like to address the core of the book that proposes vast unemployment due to automation.

Labour sources

Ford claims a great amount of labour will be freed up in the coming decades thanks to automation, I will not dwell on this since I think we can agree. The largest places for job cuts probably include automated transportation (Google car), integrated software (Mint, accounting software, medical scanning software…), self-service machines (grocery checkout, ATMs), and coordinated autonomous robots (warehouse clerk robots). Where I differ with Ford is that I think there are already many places these people can go to find new work and even more yet to be created. Notably, a recent McKinsey report suggests the internet has created 2.6 jobs for every job lost – not something Martin Ford would have predicted (executive summary here).

Labour sinks


I firmly believe that humans are creative animals. We much prefer to sing, dance, paint, film and express ourselves artistically than fill out Excel spreadsheets. Thus, with increasing automation, we should see increasing amounts of art. Based on the amount of art we see online, this is already happening and is eating up people’s time that has been freed by automation. We simply need a better way to finance the arts.


Second, there are a ton of infrastructure projects to be completed. Most societies, including developed ones, still lack high-speed rail, adequate sewage systems, roads, highways, buildings, housing, ports, energy production, and other essential items to a modern society. This is a massive human employment sink that can be driven by the government.


Third, far too large a portion of society lacks education or could be better educated. Stated simply, we need to allocate far more resources to the education of our children if we want them to build the robots and software Martin Ford predicts will arrive shortly. People in the software and high tech industry tend to be surrounded by educated and motivated people, they then conclude that people who are not highly educated must lack effort or discipline or both. By that logic, we would see far more children of software developers flipping burgers and MacDonalds, but we don’t.

In addition to changing the tax system, we need to change the way we allocate access to opportunity. If we do nothing, automation will drive us towards a plutocratic society with huge income inequality. Education is not just schools and teachers, it is a formative environment which values hard work and discipline. Access to higher education often means access to a network of people who are ambitious, driven and likely to succeed in life. A large part of affirmative action is giving access to these networks to people who might otherwise not have access; affirmative action goes well beyond providing access to educational institutions. If we hope to balance the odds for all members of society, we need a lot more people working in this field.

Efforts by people at the Khan academy (see recent TED talk) both automate the traditional job of a teacher and allow for a wider and more efficient distribution of education, but the end result is a more educated population. We just need to ensure these people can find jobs, a common joke in Russia is that it is the only country where your cleaning lady has a PhD. Again, this becomes a question of financing, but no one questions our need for more education. As George Orwell said, “Human history becomes more and more a race between education and catastrophe.” We need to run faster.

In conclusion, while I do agree that automation will continue to accelerate at an accelerating rate and will replace millions of jobs, our biggest question as a society will be how to tax and allocate resources. Central planning does not work, but neither does a free market. Rapid and vast automation will force us to make decisions sooner than we might think.

See this other interesting rundown of the book here at the Singularity Hub.

Published on July 10, 2011