A Tight Spot: Part 3: The End of Jobs? Not So Fast.
In case you missed Part 1 and Part 2, we considered how economic conditions and generative AI have made it increasingly difficult for college and boot camp grads to find work. Fewer junior positions are available, and the positions that do exist often require two to five years of industry experience. Mass layoffs coupled with remote work mean that the competition for positions is beyond anything we’ve ever seen before.
It would seem that junior developers are going the way of the dodo. In fact, many predict that developers in general are on the endangered list and that things could get so bad that everyone’s job will go away. They believe we will soon be living in a dystopian nightmare where a handful of trillionaires control all the robots, which in turn control us.
Why Are There So Many Jobs?
These dire predictions of mass unemployment and societal collapse are not new. Historically, every disruptive technology that has brought significant automation of work and subsequent loss of jobs has been accompanied by these same predictions.
In a fascinating article published in 2015, MIT economist David Autor asks the question, “Why are there still so many jobs?”. He begins the paper with a brief recap of two such disruptive technologies: the textile industry in England in the early 19th century and the manufacturing capability in the US coming out of World War II. In both situations, the number of jobs created outstripped the number lost, and in the 1960s, right in the middle of an automation scare, unemployment fluctuated between 3.5% and 5%, which economists consider at or near full employment.
The reasons for this ability of the economy to weather these disruptive technologies, according to Autor, is why we should be optimistic about the future, even with generative AI in the equation.
Autor gives two reasons why there are still so many jobs, even in the face of generative AI:
1. The number of jobs fluctuates, but the amount of work never decreases. Anyone who has survived a layoff knows that the workload doesn’t decrease, you have the same amount of work with fewer folks. A government report in 1966 put it this way: “The basic fact is that technology eliminates jobs, not work” (Technology and the American Economy, February 1966. The constant demand for work ensures that human labor remains necessary, even as some roles fall to automation.
2. Many tasks are resistant to automation, a phenomenon known as Polanyi’s Paradox. Polanyi’s Paradox. In 1966, philosopher-scientist Michael Polanyi published The Tacit Dimension. In this little book (just over 100 pages), Polanyi considers the nature of human knowledge and concludes that much of our knowledge is tacit. Polanyi knew his mother’s face. He could not say how. He knew that a skilled and experienced doctor makes much better diagnoses than whenthat doctor first began to practice medicine. When pressed to explain how, there was no answer. They just know. Polanyi condensed these ideas into a single dictum. “We can know more than we can tell.” This is what computer scientists and artificial intelligence researchers refer to as Polanyi’s Paradox.
What if We Get to Keep Our Jobs After All?
The advent of generative AI and large language models has made a direct assault on Polanyi’s Paradox. Many of these “tacit” things that we know can now be done through large language models like ChatGPT and Claude. However, we suspect that human knowledge and thinking are more than just pattern recognition.
The fact is that there are still thousands of tasks where we “know more than we can say”. In a footnote to his paper, Autor references Moravec’s Paradox: “It is comparatively easy to make a computer exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility.” (Hans Moravec, 1988, Mind Children: The Future of Robot and Human Intelligence)
Many of those tasks are difficult or impossible, even for large language models. And many of those same tasks comprise a significant part of our current jobs, tasks that require judgment, creativity, and intuition. So maybe, just maybe, even with all the doom and gloom and dire predictions of mass unemployment and societal collapse, we will get to keep our jobs after all.
What a letdown. Kind of like when planes didn’t fall out of the sky at midnight on January 1st, 2000. We felt duped. We felt played. But more than anything, we felt relief. But that relief was only possible on the other side of the Y2K “crisis”. While we may feel some relief because we believe our jobs will not vaporize in the next six months, no one knows what’s going to happen. We won’t feel full relief until we’re safely on the other side of the AI revolution.
But will we be relieved because our livelihoods are secure, or will we just be relieved, as in, relieved of duty? Autor concludes his paper, “...if human labor is indeed rendered superfluous by automation, then our chief economic problem will be one of distribution, not of scarcity.” p.28
Next time, we’ll explore whether, regardless of the outcome, we can find a way forward.
