Will robots really steal our jobs?

February 06, 2018


By John Hawksworth, Jonathan Gillham and Richard Berriman

Automation is nothing new – machines have been replacing human workers at a gradual rate ever since the Industrial Revolution. This happened first in agriculture and skilled crafts like hand weaving, then in mass manufacturing, and in more recent decades in many clerical tasks.

But as the extra incomes generated by these technological advances have been recycled into the economy, new demands for human labour have been generated and there have generally still been plenty of jobs to go around. Indeed, the UK employment rate is currently at a record high.

Recently though there have been fears that a new generation of ‘smart machines’, fuelled by rapid advances in AI and robotics, could replace a large proportion of existing human jobs. While some new jobs would be created as in the past, the concern is that there may not be enough of these to go around, particular as the cost of smart machines falls over time and their capabilities increase.

We have just published two related pieces of research that throw additional light on these issues.

The first report looks at the impact of AI on the economy as a whole – both globally and for particular regions and countries – using a state of the art ‘computable general equilibrium’ model. The broad conclusion of this analysis is cautiously optimistic – in the long run, AI should not only make a significant contribution to GDP (around 14% globally by 2030 or around 10% for the UK), but should also generate enough new jobs to broadly offset the job losses associated with automation.

Indeed, with a larger economy, the analysis suggests that the net effects on employment could even be positive in the long run, although this depends on workers displaced in some sectors being able to retrain (and perhaps relocate) to work in other sectors.

In our second study, we drill down further into the areas where jobs might potentially be automated over the next two decades. To do this, we analysed the PIAAC dataset compiled by the OECD, which looks in detail at the types of tasks and skills involved in the jobs of over 200,000 workers across 29 countries (27 from the OECD plus Singapore and Russia). We identified how this process might unfold over the period to the 2030s in three overlapping waves:

  1. Algorithm wave: focused on automation of simple computational tasks and analysis of structured data in areas like finance, information and communications – this is already well underway.
  2. Augmentation wave: focused on automation of repeatable tasks such as filling in forms, communicating and exchanging information through dynamic technological support, and statistical analysis of unstructured data in semi-controlled environments such as aerial drones and robots in warehouses – this is already underway, but may not come to full maturity until the late 2020s.
  3. Autonomy wave: focused on automation of manual labour and physical dexterity, and problem-solving in dynamic real world situations that require taking responsive actions (e.g. driverless vehicles) – these technologies are already under development, but may only come to full maturity on an economy-wide scale during the 2030s.

Our estimates are based primarily on the technical feasibility of automation, so in practice the actual extent of automation may be less due to a variety of economic, legal, regulatory and organisational constraints. But they do give a reasonable indication of the relative exposure of different industry sectors and types of workers to automation.

Which industry sectors will be most affected?

We see significant variations in potential automation rates between industry sectors depending on the types of skills required in each, although the pattern here varies over time as Figure 1 below illustrates for selected sectors. These results are based on median values across the 29 countries, although we see similar patterns in the UK, where estimated automation risks are close to median levels[i].

Figure 1: Estimated potential automation rates for selected industry sectors


Transport stands out as a sector with particularly high longer term automation potential as driverless vehicles roll out at scale across economies, but this will be most evident in our third wave of autonomous automation. In the shorter term, the financial services could see larger impacts as algorithms outperform humans in an ever wider range of tasks involving pure data analysis.

Which types of workers will be most affected?

Our analysis also highlights significant differences across types of workers and these will also vary across our three waves of automation. Figure 2 below illustrates this for people with different education levels.

Figure 2: Estimated potential automation rates by education level of workers


In the short term, as shown by the top set of bars in Figure 2, we don’t expect large numbers of jobs to be automated because it will take time for these technologies to mature and roll out across the economy. But as we move into the second and third waves of automation in the late 2020s and 2030s, we find much higher potential rates of automation for less well educated workers.  

We find less marked differences by age group, although some older workers could find it relatively harder to adapt and retrain than younger cohorts. This may apply particularly to less well-educated men as we move into our third wave of autonomous automation in areas like driverless cars and other types of manual labour in factories, warehouses and on construction sites that have a relatively high proportion of male workers at present. But, as Figure 3 below shows, female workers could be harder hit in early waves of automation that apply, for example, to clerical roles.

Figure 3: Comparison of potential automation rates for male and female workers


Spreading the benefits by boosting digital skills and lifelong learning

Our analysis highlights the need for increased public and private investment in education and skills to help people adapt to technological change throughout their careers. While increased training in digital skills and STEM subjects is one important element of this, it will also require retraining of, for example, drivers and factory and warehouse workers to take jobs in sectors like health and personal services where demand is high but automation is less easy due to the importance of social skills and the human touch.

Governments, business, trade unions and other organisations all need to play their part here in helping people to adapt to these new technologies and promoting a culture of lifelong learning. Only in this way can the great potential benefits from advances in AI and robotics be spread more broadly across society.

For more detail on how chief executives see new technologies impacting their businesses, see our latest Global CEO survey here.

[i] In general we find lower than average exposures in countries like Japan, South Korea, Singapore and the Nordic economies that have relatively high average education and skills levels. Eastern European countries, which tend to have a higher share of employment in industrial sectors that are relatively easy to automate, may see above average levels of automation according to our estimates.