The Paradox of Productivity, Part II - The Player Piano Society

AI and the Future of Professional Employment
In my last post we learned that from 1980 to 2025, technology improved the productivity of employees, but reduced the number of good jobs and workers’ earning power.  Now, in 2026, people who are knowledgeable about the state of artificial intelligence are warning that worker replacement is about to happen on an unprecedented scale.  

This image of a robot playing a piano was generated by AI.  I'm aware of the irony here.

Writing in Fortune, Matt Shumer wrote of his own experience as an AI developer. 

“For years, AI had been improving steadily. Then in 2025, new techniques for building these models unlocked a much faster pace of progress. This year, something clicked. Not like a light switch… more like the moment you realize the water has been rising around you and is now at your chest.I am no longer needed for the actual technical work of my job….
The experience that tech workers have had over the past year, of watching AI go from “helpful tool” to “does my job better than I do”, is the experience everyone else is about to have. Law, finance, medicine, accounting, consulting, writing, design, analysis, customer service. Not in 10 years. The people building these systems say one to five years. Some say less.”  

A friend on social media wrote a very similar post within a day or two of the article above.  

"I piloted the expensive version of Anthropic's Claude this weekend and we're really screwed. Anyone pretending this can't replace the majority of mid-range paid work is living in a dangerous fantasy right now.
Careers on life support:
- Product Owner
- QA/test Engineer 
- Paralegal
- Data/BI Analyst 
- Copywriter  
- Content Editor
- Any role in IT Support
- Marketing whatever 
We're not mature enough a society to have this many people out of work"

Global spending on artificial intelligence in 2025 was about $1.5 trillion, increasing at a rate of about 50% per year.  These investments are largely justified on expected cost savings for businesses by replacing employees.  But at some point, I wonder what will happen to consumer spending when many good-paying jobs have disappeared.  

The economy is already showing signs of stress for lower-income Americans.  In 2025, overall consumer spending increased by 2.7%, a slight slowdown from the previous year, and declining in the fourth quarter. Lower-income consumers faced tighter budgets and rising debt.  Credit-card debt increased faster than inflation, at 5.5%.  Delinquencies on most categories of debt rose through 2025, and the personal savings rate has declined from 6% to 4% over the past two years. 

Impact on the Economy
As I explored in an earlier post about the book “Abundance”, many workers have been left behind in the economic development of the past fifty years.  Moody’s Analytics recently reported that households with the highest 10% of income account for 50% of consumer spending.  It is the logical response of the economy to wealth inequality.  Extreme wealth inequality has existed before; e.g., in the 19th century Gilded Age, or the pre-revolutionary French economy.  I do not know how our current disproportionate economic production compares to those early ages.  But if AI greatly increases the rate of destruction of white-collar jobs, there will be even stronger division between the haves and have-nots in America.

In recent months, the expectation of what AI can do has had a major impact on the job market and in stock valuations.  The job market for software developers rapidly surged immediately post-pandemic, and equally quickly collapsed, as the capabilities of AI became better known.  Stock valuations for traditional software makers, certain internet services, and employment agencies fell abruptly.  Yesterday, the leading AI company Anthropic made an announcement about coding abilities of its Claude AI, and the IBM immediately lost $32 billion in market capitalization.

 

In my economics classes, the professors confidently predicted that new technologies would always produce new jobs.  But that is simply an empirical observation about previous economic change.  It doesn’t necessarily hold for the future.  AI has the potential for faster disruption and more complete disruption than earlier technical revolutions.  And the people put out of work by AI will not necessarily have the skills to take new jobs, if they appear.  

I’m not a Luddite.  I see no point of continuing jobs after they have been rendered obsolete.  The history of railroad employment is an example.  In 1920, there were about 2 million railway workers in the U.S.  By 1959, the number had declined to 780,000, and today, only 153,000 workers are needed to run our trains.  Around 1960, labor contracts forced railroads to continue employing workers that were unneeded due to more modern train technology, such as coal shovelers on diesel trains.  The unnecessary workers cost the railroad industry about $500 million per year, or about 5% of total profits.  

It seems to me that AI is necessarily derivative from training sets of human-generated content.  As such, AI is simply not capable of genuine creativity and innovation.  AI does have the ability to synthesize, and make connections between bits of information that a human would miss.  But, although I am no expert about AI, it seems that AI will probably not recognize new phenomena, or invent new processes for businesses.  And without creativity, technology may stagnate.  Or, perhaps AI, acting as an assistant, may give more scope to  human inventors.  Time will tell.

Artificial intelligence is necessarily derivative from training sets of human origins.  While AI may be capable of bringing ideas together in synergy, AI itself is likely to lag behind humans in terms of innovation and creativity.  

But I do believe the providing good jobs is part of a social contract between businesses and society.  Business profits should be measured as a function of how many well-paying jobs the business provides and taxed accordingly.  If the business provides well-paying jobs, great; it is fulfilling its social obligation.  If not, government can use the taxes for employment-directed programs for the public good or for job-training.  

I recently saw a conversation thread on the Reddit website.  A fellow working as a Lyft driver argued that new technology always creates new jobs.  It occurred to me that perhaps he could have had a higher-paying job if jobs had not been eliminated by new technology. Further, the job of a Lyft driver might be one of the first jobs eliminated by AI.  It just seemed to me that he hadn’t thought about the problem all the way through.

If the replacement of well-paying jobs is as extensive as expected, what will happen to consumer spending?  As noted above, half of the economy is already serving the top 10% of households by income.  What will companies make and sell when the bulk of consumers cannot afford to buy products?

 A player piano is often used to symbolize the replacement of humans by machines, as seen in the introductory title sequence in Westworld.  Kurt Vonnegut’s first novel, “Player Piano” envisioned a world where machines had replaced nearly all human labor.  Rather than a utopia, it was a dystopia, with 98% unemployment, and only a lucky few able to find meaningful work.  Wealth distribution was greatly unequal, and did not allow most citizens to engage in uneconomic but enjoyable pursuits.  It remains to be seen whether AI will create a society with better employment and more equitable wealth distribution, or whether we will have some version of the Player Piano society.  

References
https://www.bls.gov/emp/tables/employment-by-major-industry-sector.htm
https://www.bls.gov/emp/tables/real-gdp-major-demand-category.htm
https://fortune.com/2026/02/11/something-big-is-happening-ai-february-2020-moment-matt-shumer
https://www.marketplace.org/story/2025/09/17/top-10-of-earners-make-up-half-of-us-retail-spending
Player Piano, Kurt Vonnegut, 1952, 353p.