About two and a half years ago, a video called ‘Humans Need Not Apply’ was released by YouTuber and podcaster CGP Grey. In his mini-documentary, Grey talks about the increasing usefulness of robots in the society and how they are becoming more integral and necessary in the whole economic and social structure of our society. His central idea is that this continuous integration of automated machines in our society will ultimately lead to a world where human labour is no longer required to sustain systems that were inherently man-made. I highly suggest that you give it a quick watch before going through this article, as it is quite thought-provoking.
A paper by Carl Benedikt Frey and Michael A. Osborne from 2014, that seems to be a popular reference when talking about robotic automation of labour, makes a claim that is rather scary. They are estimating that about 47 percent of total US employment is at risk, showing that wages and educational attainment correlate negatively with an occupation’s probability of computerisation. The latter seems to hint that the fields that are the most likely to be affected by automated labour are the low-skilled ones—which should come as no surprise, as those are most often easier to solve for robots. However, there still seems to be a strong sense of doubt in the minds of people. It’s very easy to think “it’s been said before, but it never happened, and probably never will,” or “newer jobs will always be created,” with all the past claims of complete automation of labour falling flat.
In reality, we are at the point where human performance and machine performance are getting extremely close to one another, faster than ever. Considering that machines are performing better at an exponential rate, humans are generally getting better logarithmically—their marginal performance is diminishing, for you microeconomics students. This whole situation leaves two questions to ask:
Is my job safe?
Is anybody’s job safe?
For the first question, it is likely best explained by the example of Luddites. The Luddites was a group of English textile workers who violently destroyed power looms because they ‘feared’ that these machines, operated by unskilled workers, would take their skilled jobs. As per the Luddites, the term ‘Luddite fallacy’ is used to describe people’s concerns about long-term unemployment resulted by technological advancements—as they don’t account for the ‘compensation effect,’ they commit a fallacy. It is usually used by people who oppose their idea, and believe that in the end the unemployment effect is nonexistent and wages will increase with the overall increase in welfare due to progress.
Unfortunately, the bad news is that it’s not just the Luddites who are worried anymore, and it is less and less like a fallacy now. So, no, if you’re reading this, your job is probably not as safe as you think it is. Although, keep in mind that smashing the computer-who-took-your-job isn’t going to save your job, as it didn’t save the Luddites’.
For the second question, it gets a bit more interesting. Of course, there are a lot of jobs that are safe from robotic automation, but then most of those are subject to high competition at an international level. It’s probably a very long time until robots start programming other robots, so you need software engineers to work on developing the robots that automate work; but then those engineers’ problem is that unlike barbers, they are competing with people across the globe for their job. Regardless, I’ll be focusing on one occupational field that hasn’t gotten much mainstream attention regarding the technological developments: agriculture. Interestingly enough, the immense complexity of the variables involved in farming makes it very difficult to automate it successfully.
Problems with full automation begin with the level of attention non-staple crops require when growing them. Cereals are particularly easy to harvest because the variables your robots have to take into account are not difficult to automate the control of. However, if you want high density, and sustainable fruits and vegetables production, you need to pay extremely close attention to your soil and crops. As best told by these figures from a United States Department of Agriculture review, “Wages, salaries, and contract labor expenses represent roughly 17 percent of total variable farm costs, and as much as 40 percent of costs in labor-intensive crops such as fruits, vegetables, and nursery products,” it is significantly more difficult to work with these crops than cereals, and currently it requires a lot of human intervention and labour.
Farming is unlikely to be automated to the extent of other mass-production industries also because of the low return-on-investment of automation. Compared to the impeccably well-optimised manufacturing lines of the mega-factories you see of Ford and Intel, there are simply way too many low-hanging fruits that increase yields very significantly. For example, an Indian province was able to improve their rice yields by 45% by planting seedlings earlier in a grid pattern while keeping the soil much drier. Even though we’ve been farming crops since the beginning of human civilisation, we are still surprisingly bad at achieving efficiency in growing these crops.
Of course, comparing factories and farms is analogous to comparing laboratory experiments and battlefields. Unfortunately, Mother Nature is always going to be against your attempts at automation with the incredible levels of uncertainty she brings to your carefully-tuned processes. For strong automation, your robot needs to know about agronomy, pests, soil science, weather predictions, machinery, stock markets, and much more. Machine learning backed artificial intelligence is getting pretty good at certain stuff where it is comparable to humans, but until it is able to identify blight on tomato leaves and treat the affected plant, it has little use. Not only that, but training robots takes a lot of data and time, and although you could have a robot that could make an educated guess on, for example, when to plant your crops, a small mistake could ruin the whole season for you.
Right now, it simply looks like factories that only employ technicians becoming commonplace is much more likely than self-harvesting wheat fields or vineyards that only employ horticulturists becoming commonplace. What seems like is going to happen is robot-augmented farming, where highly specialised robots complete menial tasks like moving pots around or monitoring soil quality. Yet, this does not at all mean that people working at farms have job immunity from the merciless sickle-wielding Wheatmaster 2000…
All in all, it is important to realise that although technological change usually brings higher economic welfare, but it is not necessarily a Pareto improvement. A great chunk of the population is going to face a small increase in their overall well-being, at the cost of some labourers facing dramatic decreases in their living standards while they search for new jobs. The real need here seems to be creating flexibility in the labour market, so that when jobs are lost to robots and new jobs are ‘created,’ people can switch between them faster. Despite the role of skill in here, certain government policies like Denmark’s flexible job security one can help a lot. Ultimately, whether or not you should worry just depends on a lot of things, and the best way to make sure you are never out of work is diversifying your skill set just like you would a portfolio.
Or, you know, switch your major to horticulture—it doesn’t look like food is going out of fashion anytime soon.