AI Wealth Truth (62): Why Capital Still Wins in "Human-AI Collaboration"

I. "AI will not replace people; it will collaborate with people." "Humans + AI will be stronger than pure AI or pure humans." This is the mainstream narrative about AI. It may be true. But the question is: who benefits the most from this collaboration?

II. Let us do a simple thought experiment:

III. Suppose you are a designer. In the past, you could finish 2 design drafts a day. Now with AI assistance, you can finish 10 a day. Your productivity rises 5x. Will your salary rise 5x?

IV. Almost certainly not. You might get a raise, but maybe 20% to 50%. Where does the rest of the upside go?

V. It goes to company profit, that is, returns to capital. With the same wage expense, the company gets 5x output. The extra value goes to shareholders. This is the classic problem of surplus value distribution.

VI. Let us understand the mechanism behind it:

VII. Why do wages not grow in proportion to productivity? Because wages are determined by the labor market, not by the value you create. If someone is willing to do your job for a 20% raise, you can only get 20%. The labor market is a competitive market. Your bargaining power depends on the cost of replacing you, not the value you create.

VIII. AI lowers the replacement cost. If AI lets every designer do 10x the work, then the number of designers a company needs may become one fifth of before. Designers become easier to replace. Lower replacement cost = lower bargaining power = capped wage growth.

IX. Has history seen something similar?

X. The textile industrial revolution in the 18th century. Spinning machines increased a worker's output by dozens of times. Did wages rise? In the early stage they actually fell. Because machines let children and women operate them too, expanding labor supply. Wages only began to recover decades later, mainly through unions and labor-law protection. Productivity gains do not automatically translate into wage gains.

XI. The computer revolution since the 1970s. Computers greatly improved office efficiency. But since the 1970s, real wages in the US have been almost stagnant. Productivity doubled, wages did not. The incremental value was captured by capital.

XII. Will the era of human-AI collaboration be any different?

XIII. Technically, there is no essential difference. AI is a production tool. The return on tools belongs to the tool owners, that is, capital. Most of the value you create with AI will not belong to you.

XIV. It may be worse. AI learns faster than humans. Today you still have the ability to collaborate with AI. Tomorrow AI may no longer need your collaboration. Your collaboration value may be temporary.

XV. There is another, more hidden problem:

XVI. AI raises the performance bar. When you produce 10x with AI, that becomes the new "standard." In the past, 2 designs a day was competent. Now 10 is competent. You did not get a raise, but your workload effectively increased. You are forced to work harder just to maintain the same income.

XVII. How do companies use this?

XVIII. "AI enablement" becomes a cost-cutting tool. Companies adopt AI tools and lay off some employees. The remaining employees produce more per person, but wages do not change. Costs fall. Profits rise. This is the common playbook of tech companies over the past three years.

XIX. How do employees respond?

XX. In the short term: there is not much you can do. You can choose not to use AI, but you will be outcompeted by those who do. You have to use AI to stay competitive. You are forced into this game.

XXI. In the long term: become a capital holder. If the productivity upside goes to capital, you need to own capital. Invest in stocks, especially companies that benefit from AI. Do not only sell labor; become part of capital.

XXII. Change the rules of the game. Historically, labor rights were won collectively: unions, labor laws, minimum wages. The AI era may need new institutional design: taxing AI, universal basic income, data compensation. But these require political will, and politics is influenced by capital.

XXIII. "Human-AI collaboration" sounds like a win-win for humans and machines. In reality, the winner is whoever owns the machine. You are human, but you do not own the machine.

XXIV. Isn't higher productivity a good thing? Overall, yes. Society produces more. But distribution determines who benefits. If all the upside goes to capital, workers' situation will not improve. This is the core political economy issue of the AI era. It is not what AI can or cannot do. It is how the value created by AI is distributed. The current distribution rules are unfavorable to workers. These rules will not change automatically.