Will robots take all our jobs? Unlikely, says economist Andrew Scott

What’s wrong with the debate about whether or not “robots will take our jobs?”

If you think about how technology is going to impact the labour market, there’s loads of dimensions and the one everyone looks at is jobs. The concern is: “Will technology take all our jobs?” and I think that’s unlikely. I think it’s unlikely because that’s never happened before with technology. It’s just changed what jobs you do. My hunch is that even though there may be an increase in unemployment this time – you can’t rule that out – the biggest effect for most people will be a change in how they work. It will change the skills, the tasks and the job that you have to do.

How has technology tended to change the nature of work?

In economics there’s a sort of standard way of thinking about this at the moment; there are three channels through which technology works. There’s the displacement effect, so the machine just does my job, I’m not needed. Then there’s the productivity effect, which is that the machine makes it more productive, the firm wants more of me. Then the third one is the creation effect; jobs are created. The displacement effect is always negative, the productivity effect is always positive and the creation one is positive, so it all depends on the relative magnitude of those. Now, most of the discussion around robots focuses on displacement and it says there’s going to be massive displacement.

An important distinction is now being drawn between jobs and tasks. A job is made up of many tasks. If your job is made up entirely of jobs that can be automated then you’ll lose your job. But most people’s jobs are made up of a mixture of tasks and what happens is your job expands into the non-automated realm. A nice statistic is that in the US there are [approximately] 270 categories of jobs and only one of them has been displaced because of technology over the last 50 years and that’s elevator operator! Could it be different this time? Yeah, this is a different technology; we’re combining smarts with machines. But if you look at the history of technology, it makes us more productive. That doesn’t need to lead to more unemployment; it leads to a mixture of higher wages and shorter weeks. So hopefully that’s what we’ll see.

Could we see a spike of technological unemployment in the near future?

In general what you find is that you don’t get spikes of unemployment with technology. The worry this time is that with Moore’s Law it’ll happen [too] quickly. In the next three years, computing power will quadruple, so we’ll see the equivalent of going from nothing to driverless cars times four. Now, if that does happen, I think you will see a spike of unemployment. The only thing is, I’ve been talking about this for the last three years and I still don’t see driverless cars taking over. So I think the technology may well be exponential, but the implementation isn’t. We’ve had driverless plans for 30-40 years and we still have pilots. But again, this is all open to debate. You can see why this time it might be different.

How can we take advantage of the productivity effect?

The first jobs to be replaced were non-cognitive routine jobs. Now we’re seeing cognitive routine jobs being replaced by machine learning algorithms and cognitive non-routine jobs. That leaves cognitive non-routine jobs, and that’s the area where everyone says you want to get into, and that’s how you try to preserve your job. If you think about what machines are good at and what people are good at, the more skilled you are, the more of a chance you’ve got of succeeding. Skills that support [the digital sector] are going to be very useful, and skills [relating to] decision making, dealing with pressure, working in teams, those sort of very human skills are important.

One of the common responses to all this is to stress STEM and there’s definitely a role in that area, but I don’t think that’s how the majority of jobs will come. My worry with STEM is you already see a large proportion of STEM students don’t get jobs in STEM. I think actually the most important thing that science can give education is that very basic scientific skill of: how do I find my hypothesis, how do I test that hypothesis, how do I learn from success or failure and then develop. That’s a brilliant skill, but I think when most people think of STEM they’re talking about pretty passive rote learning and technical skills and I think that’s not what’s going to be a big area of growth. The focus on creativity and innovation is what’s really important.

How well are we coping with the impact of technology on work so far?

The sort of thing we’re most worried about in the job markets is that there are a lot of middle-class jobs that provide a reasonable income threatened by AI; the cognitive routine jobs and that whole sort of swathe of middle management jobs that I think could be automated. I think that’s the big threat. I do worry that you’re going to see the number of people doing these jobs diminish and the people still doing it won’t see their wages increase very much.

Are there any groups particularly badly affected by the changing nature of work?

If you look at the US, there’s a group that is really struggling: white non-Hispanic [men] with no college education. Life expectancy is plummeting; suicide rates are high; this is the group that’s been hit by opiates; the group that voted for Trump. If you look at the data, you’ll be aware that females now account for the majority of graduates and if you’re a low-skilled blue-collar worker whose identity is based on work and bringing in money, life is going to be very hard. At the same time, you’re seeing social changes towards equality. The role of the patriarch is no longer available, so these people are asking: what is my social role? That’s really difficult to deal with and I’m not saying this is a group that needs lots of attention and care, but there’s no longer a job for life, they’re not getting the social esteem, they’re not bringing the money in and finding a social role is actually very tricky. As the patriarchy starts to [lessen] there are also these social challenges with certain groups.

Does this require a change in social values?

Yes, but of course that’s going to be very hard, because there is also a sense that a strong back and a good character will give you a good life and that’s not the case in the world of AI and robotics. Will someone who’s got a strong back and a good character, who expected to get a decent job working in a factory, be prepared to go and work in an old person’s home? Will they think of those caring skills and that caring role as sufficiently nourishing to their social character? I hope they will, but there’s a lot of social conditioning to unwind. This isn’t just blue-collar workers: this is for white-collar workers as well.

How could that happen?

I think it’s interesting how the whole Hillary Clinton [candidacy] played out in the US, because people found the middle classes telling people how to behave somewhat patronising. So I think the solution has to come from within the community. But who is that person, what is that organisation?

I arranged an event recently on longevity with scientists and social scientists and by the end of the day I was thinking that it’s probably easier to crack the scientific problems than the social problems. To overcome people’s attitudes towards age and ageing is a very hard thing. I think it’s the same with AI and robotics. We have these concepts of work, jobs and labour and capital and they’re all out of date.