Disruption of the labour market. Part 6 of the series ‘Perspectives on Artificial Intelligence’.

Futures Studies

Of all the current technological developments, artificial intelligence is both the most profound and the least understood. We are witnessing impressive new applications, but can hardly foresee their impact on people, organisations and society. In this series of blogs – Perspectives on Artificial Intelligence – we investigate not only the opportunities, but also the intended and unintended consequences.

AI does not only affect unskilled labour

Many people think it is mainly unskilled labour that will be taken over by robots and artificial intelligence; as working activities become more complicated, they become less easy to automate. This is a persistent misunderstanding. The highly educated mistakenly feel safe.

For artificial intelligence, the number of years of study that people need to learn a particular subject is not a relevant criterion. All activities of routine character can be carried out by AI. Even in high paid jobs, for which many years of study and academic titles are required, many routine activities can often be found. And the more routine a task, the more data is made available and the higher the chance that artificial intelligence can learn to do it.

As we have already explained in an earlier blog, modern artificial intelligence systems are not programmed, but self-learn through the application of neural networks. Using examples, they identify trends and patterns in the data. Accuracy increases as the machine becomes more experienced and receives more feedback. This can happen very quickly, because when 100 people use the same AI system, the system learns from 100 different places at the same time, combining all these experiences for the benefit of everyone.

Vulnerable occupations: from factory worker to lawyer

We witnessed the first robots take over human tasks along the conveyor belt in car factories. This robotisation not only saves on labour costs, but also promotes quality, because robots respond in exactly the same way every single time. In comparison to current standards, these robot types are relatively stupid. They do not react to their environment and are not self-learning.

Modern robots that can do so, thanks to artificial intelligence, affect a wider range of occupations. Self-driving cars, trucks and buses make human drivers redundant. The work of delivery service staff can be taken over by drones, which Amazon is currently experimenting with. Domino’s is working on a self-driving pizza deliverer, the DRU (Domino’s Robotic Unit).

Future customer services will, to a large extent, be staffed by chatbots as we saw in part 2 of this blog series. You can now find more than eleven thousand chatbots on Facebook Messenger. Chatbot technology was first developed for English speakers but is now implemented in less commonly used languages such as Dutch. KLM is a forerunner in foreign language chatbots and currently offers customer services in 9 languages via Messenger.

Various applications running on the IBM Watson platform are radically changing the role of highly skilled professionals. Watson provides an expert system for doctors way beyond human capabilities. Watson analyses everything that has been published within the field of oncology and, based upon this research, provides doctors with a diagnosis and treatment plan for individual patients, as well as giving the expected chance of success. In a study in the United States, 1018 patient records were submitted to both Watson and a panel of oncologists. In 96 cases the panel did not see any viable treatment options, but Watson did. An oncologist who wants to keep on top of all the recent literature would need to spend 160 hours of each week just reading. This is practically impossible, meaning AI system support is, in this case, indispensable.

A similar application called Ross is currently under development for use within the legal profession. In the United States, more than 20 law firms now work with Ross, who/which can take on research work. When a new case comes in, Ross indicates which articles of law are applicable and what the expected outcome is based on jurisprudence. What once meant many hours of work and required significant legal knowledge has now become fully automated.

Is mass unemployment on the horizon?

In the paragraphs above we have only mentioned a small selection of professions experiencing dramatic change thanks to AI. But this is sufficient to raise a question that is of great concern to many politicians: will artificial intelligence lead to mass unemployment?

Recent and often quoted research by McKinsey shows that in developed economies approximately 25% of jobs will be replaced by automation in the years preceding 2030. Globally, this will affect 400 million people.

The fear of mass unemployment as a result of technological development is understandable, but if history is anything to go by, these fears are unfounded. When the vast majority of people worked on the land and mechanised agriculture was introduced, the same questions were asked. Instead an industrial sector was created in which many people found new work. And when factories became more efficient and production moved to low-wage countries, the services sector grew back home. Many jobs are being created in new sectors today. To give an example, more people now work in the sustainable energy sector than in the fossil fuel sector. There is a good chance that if we look back in 25 years’ time, we will come to the conclusion that both the economy and employment are doing well.

That is the macro perspective. This will not be the view of the person whose job has been robotised, and has subsequently become unemployed and no longer feels able to start a new career. Not everyone has the appropriate adaptive and learning abilities to make the switch to the new economy. At the micro level, technological development is creating victims and this requires appropriate measures.

How do we manage this transition?

A job that has disappeared due to technological progress will never come back. But we can create new work. We not only need people to develop technology, we also need people to implement it. Consider the redesign of processes within organisations. New jobs are also being created which deal with the interaction between people and robots. People need to learn how to manage and collaborate with robots and vice versa.

This is a huge educational challenge; training people for work that does not yet exist. We will return to this in a later blog. Labour mobility is on the increase, while the labour market is becoming increasingly flexible. Artificial intelligence will help to make the right match between jobs and candidates.

Furthermore, the discussion about the universal basic income is on everyone’s lips once again, as this might be an important means to help those people who have lost their jobs to find new work. Experiments are being carried out in various countries such as Finland, where 2000 people have received a basic income since the beginning of 2017. Comparable experiments in the Netherlands come with so many conditions, such as the requirement to look for work and attend interviews, that these initiatives cannot be called a universal basic income. Rather than having a debate driven by dogmas, it would be better to do proper research to find out whether a basic income will make people lazy or simply more creative.

Finally, businesses have a responsibility to anticipate technological developments. A timely and controlled transition is always better, both for the continuity of the firm and for its employees.