If humans were locked out of the 80% AI run financial markets, would they continue to function?

My second follow up question to Google Gemini:
If humans were locked out of the 80% AI run financial markets, would they continue to function?

Gemini responded thus:

This is a fascinating thought experiment, and the answer is: no, the markets would not continue to function normally, or for long, if humans were completely locked out of financial market activity.

While a significant portion of trading is automated by algorithms and AI, they are still fundamentally designed and operated within a human-defined framework and rely on human interaction at various levels. Here’s why removing humans entirely would lead to a breakdown:

  1. Purpose and “Why”:
    • AI/Algorithms lack intrinsic goals: Financial markets exist to facilitate capital allocation, investment, and risk management for human needs and goals (e.g., retirement, company expansion, wealth creation). Algorithms and AI don’t have these motivations. They are designed to execute strategies based on parameters set by humans.
    • No “real economy” connection: Markets reflect the real economy – companies producing goods and services, people earning wages, spending, saving, investing. If humans are removed, the fundamental economic activity that gives the market its meaning and input would cease to exist.
  2. Human Input and Oversight are Crucial:
    • Algorithm Design and Updates: Humans design, test, optimize, and update the algorithms and AI models. Without human programmers, quants, and data scientists, these systems wouldn’t evolve, adapt to unforeseen circumstances, or be maintained. Bugs wouldn’t be fixed, and new strategies wouldn’t be developed.
    • Risk Management and Circuit Breakers: While algorithms can have built-in risk parameters, humans ultimately define the overarching risk tolerance and put in place “circuit breakers” or emergency shutdowns. In extreme, unprecedented market events (like a “flash crash” potentially triggered by automated trading), human intervention is often needed to stabilize the system.
    • Interpretation of Unforeseen Events (“Black Swans”): AI and algorithms are trained on historical data. They excel at recognizing patterns in known situations. However, “black swan” events – truly novel, unpredictable occurrences (e.g., a sudden geopolitical crisis of a type never seen before, a pandemic with unprecedented economic effects) – are difficult for AI to interpret and react to appropriately without human judgment. Their reaction might be to simply shut down or exacerbate the problem.
    • Setting Parameters and Objectives: Humans set the “rules of the game.” What are the market hours? What are the trading rules? What constitutes a valid transaction? All these fundamental parameters are human-defined.
  3. Liquidity and “The Other Side of the Trade”:
    • Who is Buying/Selling? If all human participants (retail investors, institutional investors, pension funds, hedge fund managers, etc.) are removed, who would be placing orders that the algorithms would execute? The market needs buyers and sellers. While algorithms could technically trade with each other, they are ultimately trading on behalf of human-owned entities or with the expectation of future human interaction.
    • Capital Flows: Capital ultimately originates from human wealth and economic activity. If there are no humans to generate wealth, save, or invest, the capital pool would eventually stagnate and deplete.
  4. Trust and Confidence:
    • Financial markets operate on trust and confidence. If humans were locked out, who would trust the system? Who would trust that the underlying companies still exist or that the financial system is still sound? A breakdown of trust would lead to a complete withdrawal of any remaining human capital.
  5. Regulatory and Legal Frameworks:
    • Markets are governed by complex legal and regulatory frameworks, all created and enforced by humans. Without human oversight, these systems would cease to function, leading to chaos and potential fraud.

In essence, while AI and algorithms are incredibly powerful tools for executing trades and managing data, they are not the raison d’être of the financial markets. They are instruments used by humans to achieve human economic objectives. Removing humans entirely would be like having an incredibly sophisticated automated factory with no one to design the products, supply the raw materials, or buy the finished goods. The machinery might run for a bit, but it would quickly become meaningless and grind to a halt.

How much of the financial market can we trust AI with?