Our understanding of the potential impact of automation is undermined by a lack of differentiation between jobs and work. Robots can do work but no robot is a direct replacement for a human employee.
There is one characteristic that is more important than any other in humanising the robots we see in science fiction films. It is not the oversized eyes of Wall-e, or the cute little beeps of R2-D2, or the conversational humour of Johnny 5. It is their ability to adapt to different challenges and situations.
This adaptability is what separates the robots of science fiction from the robots of reality. In the real world, we tend to build robots for single tasks. We design them to perform those tasks with incredible efficiency. Inside the bounds of these narrowly defined tasks, they can outperform humans by many orders of magnitude in speed, strength, and dexterity. Beyond these narrowly defined tasks though, they are useless. Unless and until they can be reconfigured for the next challenge.
Building flexibility into robots is expensive, both mechanically and computationally. This is why the robots of science fiction are so different and so appealing. R2 can deliver cocktails, hack space stations, fix your space fighter, and even hold down a conversation, if you speak robot whistle. Meanwhile his real-world counterpart can just move steel pressings from one place to another, over and over again.
Robots < humans
This differentiation is critical in understanding the impact of robots on the future workforce. Because there is no robot today that is a straight swap for a human being. When we employ a human in a job, we count on a degree of flexibility and an array of complementary skills. They must be able to understand variations in the brief and respond to them. They must be able to access the task location. They must be able to communicate with the other human beings around that task: customers, colleagues, partners.
We do think about these things when we recruit a person for a role. But a lot of the basic skills they require are just part of being human. They are not part of being a robot.
So, when robots enter the workforce, while they absolutely displace people it is never a one-to-one ratio. Rather, the workload of specific tasks is aggregated from multiple people and allocated to one robot. One machine might do 80% of the work of ten different people. That still leaves work for two full time people, but it is now a pair of roles that look very different to before.
Sometimes those new roles will be very high value. For example, in a professional services environment, the robot might do a lot of the document processing that has traditionally been the domain of junior members of staff. What is left is the strategic thinking, problem solving and client engagement.
Sometimes what is left might be less engaging. Picture the delivery driver in a self-driving van who just has to get out at the relevant locations and run the parcel to the door (or as is so often the case, throw it over a fence or put it in the paper recycling bin).
Of course, you don’t have to use a human to round-out the robot’s capabilities. You can always change the operating model. In the delivery example, the robot truck is more likely to park up outside and phone the recipient to tell them to come out collect their parcel. It won’t need to put things in odd places because it is in permanent communication with the recipient’s smart device and knows their location and availability.
High value, low volume
The net result is likely to be that the perceived value of humans in the full-time workforce increases. Because humans do the low volume, high value tasks that machines find difficult. This won’t, sadly, overcome our historical underpayment of those in roles like care and teaching. Though the same automation effects may free more of their time to focus on the aspects of their job where they add the most value and relieve some of the time pressure.
Alongside the rising perceived value of human work, it is hard to see anything but a decline in the total number of traditional full-time jobs. Jobs that are a mutual contract between employer and employee, trading commitment for security and personal development. Though the statistic about ‘65% of future jobs not being invented yet’ seems to be itself, completely invented, there will undoubtedly be new jobs created in the future. But it is hard to see what jobs might be created that offer large numbers of people long term security.
This is because of the way that current generations of automation technology differ from previous generations. First, though specialised, today’s robots absolutely can take on cognitive tasks. Second, the cost of tailoring a machine to a specific task is much lower than it has been – especially where those tasks are cognitive and can be replicated in software. It’s hard to see new work being created that can’t itself be undertaken 80% by robots.
Widening the divide
The jobs that remain will be of higher value and – on average – higher pay. The people who don’t get those jobs? There will be the same spread there is now amongst the self-employed and gig economy workers. Many with high value skills will be absolutely fine. But the size of the ‘precariat’ in less secure, freelance and part-time work could grow considerably.
Robots don’t take jobs, but they do take work. And in doing so, they may widen the economic divisions in society.