Automation is a more severe threat to the global economy than people think.

Up until now, most of the tasks we’ve automated were deterministic. Numerical computation is a good example of this. Calculating the value of a math expression (like “2 + 2 * 2”) follows the same rules each time. TurboTax is also pretty much deterministic — it has a bunch of special cases for extra taxes and deductions. For deterministic problems, the input has a pattern to it, and once we know the input we know exactly what the output should be.

Conversely, other tasks have a “human” touch. The work that doctors and lawyers do is on this list because they have to deal with and persuade humans for a living. Therapists and teachers are in this area as well.

Through this classification, we have a certain conception of what computers are “good” at. When we think about the jobs most at risk of being automated, we think about accountants, traders, mathematicians, and anyone else who applies patterns and logic in order to solve problems. Doctors, lawyers, and teachers, on the other hand, don’t find themselves outclassed by computers very often.

It’s a misconception, however, to think this latter class of jobs is immune to being turned into an algorithm just because it deals in humanity and emotion rather than numbers and rules. Indeed, there is good reason to think that many human behaviours can be modelled as statistical processes that can be emulated by computer hardware. At its most basic level, the human brain is a network of neurons that is trained to respond to internal and external stimuli. Similarly, machine learning models can be trained to approximate human behaviour, or better yet, to approximate rational behaviour.

This means no job a human can do is safe from automation. Jobs that involve driving (e.g. trucking) seemed pretty safe until advances in sensors, microprocessors, and internet connectivity brought us to a point where Google, Tesla, and Uber are all competing to get their self-driving car to market first. With more precise health tracking via wearable devices, doctors’ diagnoses are being improved and negative trends are being detected more quickly. Legal advice, education, and countless other “non-deterministic” areas are being disrupted in a similar way.

Of course, most of these areas of advancement have only augmented humans so far, rather than supplanting them. Even making these jobs easier, though, will mean that companies will be able to cut staff. And it’s not hard to imagine the arrival of near-human capabilities for these technologies over the next five to ten years.

If such advances come to pass, the fundamental assumption of the global economy — that economic growth is generated by human labour — will be disturbed. I don’t know what the response should be if this happens, but basic income seems like something worth considering.

Follow Ideas and Words using the button below for more like this!