“Genius is one percent inspiration, ninety-nine percent perspiration” is attributed to Thomas Alva Edison and has some truth to it. This is somehow confirmed by the 10000-hour rule: it takes a lot of practice to master a field. The exact number may not be true, but the thought behind it certainly is. There is such a thing as talent, which makes it easier to reach a certain level faster than others, but it still requires hard work.
In all fields, it’s not about perfection of a repetitive task, but about developing some intuitive understanding of how things fit together and work together. A professional financial auditor recognizes unusual patterns in a balance sheet; an experienced programmer often understands what a particular piece of code does just by skimming it. This understanding of patterns and relationships makes them experts in their field.
Some people achieve this level of expertise; others never will, despite decades of experience. In the following discussion, I assume a normal distribution of skills and expertise in a given field. Most people will be able to develop an average level of expertise, and a few will achieve mastery. These masters are the ones who push the boundaries and advance the field.
AI and Human Expertise
Now let us take a look at the AI entering the scene. As GPT-4 and Midjourney impressively demonstrate, a lot of average knowledge and creative work can already be done by AI. Writing an agenda for a workshop? AI will do that for you. Summarize the current mainstream views on a topic? AI will do that for you. Write some code for a new service you are developing? AI can do that for you, too.
To be honest, many tasks that I would leave to a junior consultant will probably be done faster and with better results by an AI. The logical consequence will be that fewer and fewer junior roles will be needed. What we need are the experts, the people on the far right of the bell curve.
This will break the old progression from apprentice to journeyman to craftsman to master. Fewer apprentices (entry-level positions) mean there are fewer people to train as master craftsmen. Sure, master craftsmen will be in high demand and very well paid. But for those who have not reached that level, jobs will be few and poorly paid.
The Rockstar Markets
In some areas, this kind of market has been around for decades. In the (popular) music industry, a few stars at the top make a fortune, while 99% of musicians and singers, most of whom perform just for fun and as a hobby, will have a hard time. A similar situation exists in sports: while there are superstars in some sports who earn millions, the majority of competitive athletes will not be able to make a living at it. And science is not far behind: people who are dedicated to science could be in danger of dropping out of the system altogether if they are unable to hold a chair in their field after a certain age.
Given the lower demand for low to mid-level expertise in knowledge-intensive and creative professions, I would expect these fields to evolve in a similar fashion. So a few rock stars, a lot of amateurs doing something as a hobby. There has already been a long discussion about the consequences for society if the majority of knowledge-intensive and creative tasks are taken over by an AI owned by a small minority, so I will leave that topic out. What I do find interesting, however, is the question of how we will develop those masters whose skills go beyond what AI can do today.
Example: Rational Career Choice
Applying the concept of opportunity cost to rockstar markets makes the problem easier to grasp.
An example will illustrate this: Suppose Gordon has an equal chance of becoming a very skilled nurse or a master software engineer. The market for nurses will remain as it is today, the market for software engineers will turn into a rockstar market.
In the EU, the average annual salary is € 35.300,00. Assume that he has a 99% probability of earning this average annual income and a 1% probability of becoming dependent on welfare (at €12,000.00). The expected value of this occupational choice would be € 34.770,00.
| Annual Salary | Probability | Expected Value |
|---|---|---|
| 35.000,00 € | 99% | 34.650,00 € |
| 12.000,00 € | 1% | 120,00 € |
| 34.770,00 € |
If we now assume that the probability of reaching the level of Master Software Engineer in the future is 1%, Gordon will be dependent on social welfare 99% of the time (€ 12.000,00). Now, what would be the salary required to make an equivalent career choice, i.e., to have the same expected value? This is again a simple calculation that leads to an annual salary of € 2.289.000,00.
| Annual Salary | Probability | Expected Value |
|---|---|---|
| 12.000,00 € | 99% | 11.880,00 € |
| 2.289.000,00 € | 1% | 22.890,00 € |
| 34.770,00 € |
Assuming a rational decision, Gordon should choose a career as a nurse if the annual salary of a software engineer is “only” €2.000.000,00.
So how will people be motivated to pursue careers in any of the affected fields? With music and sports, there are certainly side benefits that make them worthwhile even without monetary rewards. But what about accounting, law, or developing new algorithms? Will people do that if they have to earn a living doing something else they actually enjoy? And in the end, does this mean that sooner or later all research and development in areas amenable to AI will be dominated by AI?
The AI View
I asked ChatGPT for a comment and this is the answer:
Your observations on the effects of AI on various fields and the potential reduction in entry-level roles are well-founded. As AI systems continue to improve, they will indeed take over many tasks that were traditionally done by junior employees. This could lead to a smaller pool of people available to develop into masters of their craft.
Motivation to pursue a career in an affected field will likely come from a combination of factors. Some people might still be driven by the passion for a subject, the desire to solve complex problems, or the prospect of potentially becoming one of the highly sought-after experts. Moreover, with the reduction of entry-level positions, it is possible that companies and educational institutions will adapt by offering more specialized and focused training programs to develop expertise at a faster rate.
Another possibility is that the role of human professionals will shift towards tasks that AI systems are not yet capable of handling, such as tasks requiring a deep understanding of human emotions, cultural context, or ethical considerations. In this scenario, human experts would collaborate with AI systems, leveraging the strengths of both parties to achieve better outcomes.
As AI continues to develop, it is also important to invest in AI research and development, as well as in responsible AI usage guidelines. Human expertise will be necessary to guide the development of AI systems, ensuring that they are aligned with human values and ethical principles.
It is also possible that with the increasing automation of knowledge-intensive and creative jobs, society will place a higher value on interdisciplinary skills, adaptability, and critical thinking, as these are the skills that will enable people to work effectively alongside AI systems and adapt to the changing job market.
In conclusion, the advancement of AI is likely to change the landscape of many fields, and it may indeed lead to a smaller pool of human experts. However, this does not necessarily mean that AI will dominate all exploration and development in these fields. Instead, it could lead to a future where humans and AI systems collaborate, leveraging their respective strengths, and the focus shifts towards developing skills that complement AI capabilities.
What do you think? Is ChatGPT right?