Artificial intelligence and jobs: What’s left for humanity will require uniquely human skills
Originally posted on The Globe and Mail.
When the machines took over farming, a new set of industrial jobs blossomed. When the robots took over the factories, we moved to IT jobs that had never previously existed. Now that AI is taking over another swath of jobs, a wave of as-of-yet-unheard-of jobs, will soon flourish.
Or, will it?
The thinking that leads to this conclusion has a long, decorated history going back to Joseph Schumpeter’s description of creative destruction in the 1940s. According to Schumpeter, the “gale of creative destruction” describes the “process of industrial mutation that incessantly revolutionizes the economic structure from within, incessantly destroying the old one, incessantly creating a new one.” This time around, however, he is likely wrong.
Humans do not have an infinite set of capabilities. We are capable of many things, but as demonstrated by AI systems such as Watson and AlphaGo, humans can be outperformed by machines on most individual tasks, no matter how complicated they appear at first. For years, Texas hold’em poker, with its complexities of betting strategies, luck and personalities, was seen as a great example of one of the challenges only a human could handle. This belief eventually fell when a poker-playing AI system called Libratus beat the best in the world in 2017.
There are those who would argue that these are all examples of AI outperforming humans in the “digital” world, and that real-world challenges will prove much more difficult. Each month, however, brings fresh progress in self-driving cars, drones and robots navigating urban terrain comfortably. It is becoming increasingly difficult to find any sacred areas that robots will not be able to perform better than humans.
The future of human employment can be likened to a landscape that is gradually inundated with water as water levels rise. The landscape represents the skills required for different jobs and the water level represents the capabilities of machines. The crude machines that replaced manual labour on farms would be represented by the beaches and low-lying land being flooded. When robots replaced workers in factories, the lowest plains went underwater. At each step, there was land to retreat to, and that “land” was things that the machines were unable to do — generally higher skilled, higher paid and more interesting jobs. But the hills and mountains are not infinitely high and as the water rises and mountains become islands in the rising sea, the available land grows smaller.
So where should we look for this final archipelago of human employment? The best place to start is deep within ourselves. As much as we pride ourselves on advanced skills such as mathematics and chess, humans are not born innately aware of algebra or checkmate. We are, instead, a social species. We are born innately aware of others, their reactions to us and our relationships with them. Removing a person from a social environment is so harmful that it is deemed to be a form of torture and is banned by the Geneva Convention.
When we attempt to use machines to replace the role of humans in our social lives, the response is immediate and negative. The movie industry has learned this lesson the hard way as a number of films, such as Final Fantasy and Ex Machina, have used computer graphics to generate almost realistic humans and the audience revulsion was intense. Attempts at almost-human robots that interact and demonstrate emotions are called frightening, uncanny and creepy in almost every description. Recently, Google demonstrated an eerily realistic voice interaction called Duplex that caused great offence around the internet. The reaction was a raw emotion of creepiness, disgust and a sense of fraud.
The reaction, interestingly, is not that the technology is inadequate. In fact, it is extremely capable and would be able to perform better than humans at the task in question. The point is that we, as a society and as a species, don’t want AI to replace our social interactions and our relationships. It’s a part of what makes us human and it’s a part that we intend to keep.
The last islands of human employment therefore won’t be based on skills that machines are incapable of; they will be areas where we don’t desire AI replacement. Relationships, trust, guidance, caring, nurturing and social interaction are traits that these jobs will share. Many of the underlying tasks, behind the scenes, in those roles may be automated, but the relationship cannot be, and will not be.
Relationships, based on our nature as primates, are core to what it means to be human. When we finally arrive at a future in which AI has been explored across the full extent of human employment, we will find that jobs that rely on trust and relationships are the jobs that remain.