The End Of Well-Paid Knowledge Work Jobs
The Upper Middle Class’ Reckoning Is Coming — And The World Isn’t Prepared For It
“Take a seat John”, says the HR manager in his blue suit. “There’s something you need to know.”
John is a financial analyst. After getting a Bachelor’s in finance and his MBA from a well-known (but not Ivy League) school, he got a job at the investment arm of a major American bank. His day-to-day tasks center around analyzing financial statements, market research, client meetings and the holy grail of his profession — creating financial models.
He’s 28 years old and has been working at the bank for nearly 3 years. He works in Chicago and makes $85,000 per year before bonuses.
John lives a comfortable life — he lives in a cozy one-bedroom apartment in the financial district. He eats out twice a week with friends or his girlfriend. He regularly buys $5 lattes at Starbucks against all financial reason. He travels to Miami or the Caribbean twice a year.
In other words, John is the picture of success in a 21st century developed economy.
For the past 40 years, the Western world has rewarded people like John. Smart and conscientious workers have flocked to highly skilled knowledge work jobs like financial analysts, management consultants, marketing professionals, lawyers and accountants.
For those smart or disciplined enough to study their way into these professions, the promise has always been simple: stable employment, a salary guaranteeing a ticket to the prized upper middle class, and good working conditions.
In fact, until recently, these workers have been in fairly short supply. It turns out processing complex data into useful insights is no easy task.
There’s only one problem: the story does not take place in 2023 — the year is 2026. On a windy Autumn morning in 2026 in front of his HR manager, John just got the news that he was let go from his current position.
He was offered a lower-level position as a “financial supervisor” — the term used for a new job supervising a generative AI charged with analyzing data and building financial models on its own. All he would have to do now is ensure that the AI does not make any major errors. The new pay he was offered is $45,000. He’s competing for that position with hundreds of less-educated workers hungry for any job.
A dystopian scenario? Perhaps. Yet, the current evidence around the effects of generative AI on employment is clear: we might be moving toward the “assembly line” of knowledge work. The Western world is definitely not ready for it.
The Assembly Line Of Knowledge Work: Reducing The Gap Between The Ultra-Skilled And The Unskilled
Earlier this year, Harvard Business School and the Boston Consulting Group conducted a joint study. Their aim was to study the effects of GPT-4 and generative AI on knowledge work productivity.
The headlines around it were extremely positive: the workers produced higher quality work (nearly 40% higher), they completed 12.2% more tasks and they managed to complete them 25.1% more quickly.
But the lead might have been buried. The biggest performance gains came among the less highly skilled. In other words, top performers did not benefit nearly as much from generative AI as the bottom ones. We’ll come back to the implications of that nuance in a moment.
Another, less optimistic study conducted on the effect of generative AI on online freelancers also had fascinating results. The study analyzed the number of freelance jobs (copywriters and graphic designers) and earnings on online gig platforms such as Upwork.
It showed a 2% decrease in the number of monthly jobs and a 5.2% reduction in monthly earnings on the platform following ChatGPT’s release.
Once more, the key nuance was that what the researchers deemed “high-quality workers” were disproportionately affected. The knowledge workers “cranking intellectual widgets” (i.e performing simple and extremely repeatable tasks) were less affected.
Both of these studies show that generative AI might allow knowledge work to move from a small elite class of intellectual “craftsmen and artisan” to an assembly line.
Just as Henry Ford broke down the process of making a car into dozens of separate assembly lines, generative AI might allow firms to break down the process of turning raw data to insights into many separate jobs.
Why is that? With a simple set of instructions, less-skilled workers will be able to turn existing information and data into useful insights with the right prompts. The generative AI still needs an operator — that operator just doesn’t need to be a very educated one. And each step of the “intellectual value chain” can have a single operator using pre-determined prompts.
The upside of this scenario is that it would open up knowledge work to a large share of the population currently “excluded” from knowledge work. The introduction of the assembly line was a boon for millions of uneducated workers in the early 20th century. If we fast forward the analogy to today, it might be a blessing for millions of lower-skilled workers. Yet even that hope might prove unfounded as we will see shortly.
And as usual, one class of people will benefit enormously: the capitalists and star employees.
Welcome To Utopia: Wealth Accumulation & Inequality Under Generative AI
The economic pie is always divided up between land, labor and capital. The future of land is beyond the scope of this article. What is clear is that generative AI will probably pose another devastating blow to labor.
Over the last 40 years, wage growth for most jobs in Western countries has been stagnant or declining. Globalisation flooded the labor market with cheap offshore labor to perform menial tasks.
There was always at least one silver lining: highly skilled “top 10%” jobs still showed healthy wage growth.
This was because highly educated knowledge workers were scarce and therefore had bargaining power. This allowed them to extract more “profit dollars” from corporations compared to the unskilled.
The capitalists needed them more than they needed the capitalists. Silicon Valley’s obsession with offering stock options and ping pong tables to their employees is the perfect illustration of this phenomenon.
It now seems like this picture might change soon. If generative AI can indeed reduce or eliminate the skill gap between top and bottom knowledge workers, it will also eliminate the bargaining power of highly skilled knowledge workers. That means their wages will inevitably have to decrease.
You might point out that it should be compensated by the millions of menial but steady jobs created by the new “assembly line” system of knowledge work. Yet there is no particular reason why these jobs need to be performed in Western countries. Hundreds of millions of Indian and South-East Asian workers will be happy to perform low-pay research and data entry.
What this means is that the ability for companies to create value in the economy will require even lower labor costs than ever before. This year, we’ve heard many talks of “on-shoring” due to global macroeconomic uncertainties. This might be true for physical, industrial work — but knowledge work itself seems destined to keep moving offshore. After all, there is no need for a shipping container and there are no import taxes to send instructions on Slack to a data entry worker in India.
Lower labor costs and higher labor productivity — that’s the recipe for increased corporate profits. Once more, capital seems destined to accumulate a bigger share of what will be, to be clear, a growing pie. The picture is never one-sided — as Churchill famously said, “The inherent vice of capitalism is the unequal sharing of blessings. The inherent virtue of Socialism is the equal sharing of miseries.”
Another class of people who will benefit will be the star employees or freelancers. Sure, the entry and processing of data might become commoditized. Yet it’s clear that a few stars will always be needed to create the systems and operations needed to strategize how to extract more value out of generative AI.
In fact, as I’m writing this article, OpenAI (the parent company behind GPT-4) has been reported to offer up to $10 million pay packages to attract top AI researchers.
Star software engineers, strategists and finance geniuses will become even more valuable than ever. A single employee will have access to an army of research and data entry employees along with powerful algorithms.
This means that the wages of top 1% and 0.1% workers should keep rising. Instead of the top 10% of employees benefitting, it would narrow down even further.
The framework seems simple: any task that can be taught in a straightforward series of steps is destined to be automated, assembly line style. The few jobs that create incredible value and that can’t be systematically taught are destined to prosper. They also happen to be rare and extremely difficult to master.
Conclusion: Living In A “Post Knowledge Work Boom” World
Many of our beliefs and institutions have been built around the importance of highly skilled knowledge workers. Our education system is built to push workers into these jobs. Their lessening importance might have devastating consequences for the next generation.
The picture is not all dark. If the last two industrial revolutions have shown us anything, it’s that when jobs are made redundant or offshored, the economy gets richer. That opens up new classes of jobs to serve ever-higher levels of needs on Maslow’s hierarchy of needs. The transition can be painful — think of the terrible working conditions of workers in factories during the first industrial revolution or the mass unemployment in the American rust belt during the offshoring of industrial jobs — but it usually has a positive outcome.
“Relational” jobs (involving human interaction) such as sales, account management, psychologists and coaching for example seem destined to grow rapidly. Manual labor jobs such as plumbers and construction workers might rise back to prominence. Jobs servicing the increasingly rich class of capitalists and star workers might also do well.
After all, we’re still far from having functional humanoid robots, and a cheap offshore worker cannot fix your sink. AI seems a long way off from being able to interact with humans as well as humans themselves.
Yet if smart and disciplined young people cannot get access to high-paying knowledge work jobs after their long education, the American Dream will come into question once more. Already under threat, with many young people disillusioned by their inability to buy a house, it might be the final blow to this core pillar of the Western world.
Democracies and capitalist economies function well when most people’s living standards increase over time.
How long can a democratic society keep functioning when more and more of the economic pie goes to an increasingly smaller class of stars and capitalists? Does it even matter how the pie is divided if it’s growing so quickly that everyone’s quality of life increases regardless? Only time will tell.
SOURCES
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4527336
Fantastic article, can’t wait to read more
Excellente vision, qui a le mérite de nous pousser du confortable 10% au 1%.
Dans, la droite ligne des propos de Tugan Bara, dans la francophonie..
Pour info, j'ai trouvé des dev informatique à 7$/h sur Tiktok !