3 behavioral trends for an autonomous future, part 2: Expert knowledge as a non-rival good
In the previous post in this series, we talked about how humanity is becoming accustomed to letting our consumption decisions be guided by algorithms, rather than the recommendations of friends. In this post, we will explore together a second, related behavioral trend: the increasing comfort with turning to algorithms for expert knowledge and advice.
For the modern knowledge worker attaining expert status is the pinnacle of achievement. Today, widespread general education is the norm in the developed world. For example, in the United States, some ninety percent of the adult population has graduated from high school. That’s fourteen years of schooling to aspire to a job paying perhaps minimum wage. To be truly competitive in the job market might require eighteen to twenty-three years, including university degrees and on-the job-training in the form of internships or entry-level programs. It is those last few years, spent accumulating specific knowledge and training to become an expert in a field, that make the whole difference in expected income.
Aspiring to a higher income by becoming an expert requires the sacrifice of a piece of our life. The currency we pay with is time and in many cases our originality. It seems like the goal of modern educational system, at least at the primary and secondary levels, is to homogenize thought and behavior. In those first years of schooling, we learn some civility and manners framed by a certain world view, absorbing best practices for teaming up with our companions and accomplishing tasks. It is only in later years that we add domain expertise on top of that foundational platform, with the intent to one day charge society a fee for accessing that expertise. Knowingly or not, that’s the implicit deal each individual makes with society: the more a person sacrifices in the process of learning and training, the more they expect to be able to charge for their expert contribution to society.
In many ways, the actual dynamics of learning, buying and selling expert knowledge is too inharmonious to reach an expansive use of humanity’s existing brainpower. For example, even if at the aggregate level there is a positive relationship between time spent learning and expected income, we have to keep in mind that not all experts are able to charge the same. For each person the outcome is in probabilities and many are no way near the average expected income. We cannot always blame it on poor quality of education, other variables like psychological traits and affiliation to certain human networks are also part of the equation. Another aspect is that location matters; because it takes experts to train experts, they tend to be concentrated in hubs — reinforcing relative geo-imbalances. Then after being trained, each expert has limited time capacity to interact with those seeking advice.
In the past decade, the growth of e-learning mitigated imbalances in expert training. E-learning solves the location and scheduling constraints of learning, including expert learning. Today we have platforms offering high-quality online content twenty-four hours a day and all year long from the best domain experts in the world in the most advanced fields of knowledge. Now advances in machine learning, applied to build applications in fields like investment, medicine, accounting and law, promise to create a revolution in the way society approaches the whole dynamic of learning, buying and selling expert knowledge.
The use of machine learning to create algorithms capable of providing expert advice is not new. As I’ve mentioned before, we’ve been building those machines in the quant trading world for decades. But there it was mostly experts interacting with algorithms giving expert advice. What is new in finance is that products built on expert algorithms are interacting with non-experts, and users feel comfortable adopting them. In the US, retail investors commonly use a variety of robo-investing products and services to build their investment portfolio. In medicine, machine learning is being used to help with diagnosis, monitor treatment, assist in surgery, and predict critical health events like stroke or cardiac arrest. In accounting, machine learning is being applied in auditing, to detect fraudulent expenses, and to clear invoice payment — giving new types and smaller size users access to a kind of accounting expertise they couldn’t afford before. In law, algorithms can review contracts and documents, flag the parts that need revision, and predict legal outcomes, reducing legal bills and providing access to new users for whom those services were too expensive.
As Alfred North Whitehead wrote in his book An Introduction to Mathematics, “Civilization advances by extending the number of important operations which we can perform without thinking about them.” By trans-boarding expert knowledge from algorithms running on a human brain to algorithms running on a tech platform, we are converting humanity’s knowledge base from a limited good, governed by scarcity, into what’s called a non-rival good — one that can be consumed by one person without limiting its consumption by others. This trend is accelerating and it will entirely transform the way we structure the learning, buying and selling of expert knowledge — eliminating constraints in the breadth and speed of global wealth creation
In the next post in this series, we will go over the third behavioral trend: humanity’s constant quest to expand our physiological limits.
