The Growth and Performance of Artificial Intelligence in Asset Management -- by Shuang Chen, Clemens Sialm, David X. Xu

TEXT DISSECTION

TEXT START:

"We examine the growth and performance of AI-driven investing. Using investment advisers' regulatory disclosures, labor market data, and fund strategy descriptions, we document that AI-driven investing has grown steadily since the early 2010s and is concentrated among hedge funds. AI hedge funds outperformed non-AI hedge funds in early years, but this outperformance declined over time, even among early adopters."

THE DISSECTION

This is a competitive equilibrium memo dressed as empirical research — documenting, with meticulous academic rigor, the terminal phase of a temporary alpha window. The paper accidentally confesses what its authors likely intended as a neutral empirical contribution.

What it actually documents:

The "early years outperformance → decline over time" pattern is not a story about AI limitations. It is a pure replication proof — alpha from AI adoption was arbitrage in a thin market. As adoption expanded, the alpha compressed toward zero. This is textbook efficient market convergence. The paper proves that AI-driven investing behaves exactly as DT predicts: it is a competitive displacement mechanism, not a durable skill premium.

The critical reveal: "Concentrated among hedge funds."

This is the Sovereign-Servitor split in pure institutional form. Hedge funds = capital owners / asset controllers. The paper is literally documenting the early formation of the financial services Sovereign class — those who own and deploy AI capital in markets — while saying nothing about the mass of workers displaced from analytical roles. It is the view from the penthouse. The view from the demolished trading floor is absent by design.

"Lower return comovement" is presented as a counterintuitive finding that soothes concerns about AI homogeneity. But read correctly: if AI hedge funds have lower comovement than human-run funds, this means AI is enabling more diverse strategies at scale — accelerating competitive saturation across a wider strategy space. The alpha decay will be faster and more complete than the paper's mild language suggests.

THE CORE FALLACY

The paper treats AI as a source of alpha (investment edge) rather than what it structurally is: a cost arbitrage mechanism that eliminates the labor premium of analytical work. When AI reduces the cost of financial analysis toward zero, it does not create sustainable alpha — it destroys the economic basis of the human analyst class. The alpha observed in early years was transitional rent from being early. The paper documents the rent collapsing in real time and calls this a "limitation."

It is not a limitation. It is the mechanism functioning correctly.

HIDDEN ASSUMPTIONS

1. Alpha is the right frame. The paper assumes AI's value in asset management is whether it generates investment returns beyond human managers. DT predicts the real value is eliminating the need for human managers at all — not beating them, replacing them entirely.

2. Hedge fund survival is assumed. The paper studies AI adoption within the hedge fund industry. It does not question whether hedge funds as an institution survive the displacement of their own competitive logic. Asset management firms that replace human analysts with AI are not winning — they are liquidating their own human capital base while racing competitors to do the same.

3. Labor market analysis is decorative. The paper mentions "labor market data" but the abstract reveals no findings about who loses. It documents the growth of AI without a corresponding accounting of the human displacement this growth represents. The workers are the ghost in this dataset.

4. "Outperformance declined over time" is framed as bad luck. It is actually mathematical inevitability once AI adoption is competitive. Any alpha window in a liquid market attracts capital, compresses the window, and closes it. The paper treats this as an empirical puzzle; it is a theoretical certainty.

SOCIAL FUNCTION

Prestige Signaling + Transition Management.

The paper performs academic rigor around a phenomenon that, correctly interpreted, announces the liquidation of its own subject matter. Asset management academics studying how AI replaces asset managers is like ornithologists documenting the extinction of birds — with the crucial difference that the ornithologists are not themselves birds.

The "both potential and limitations" framing is false equivalence theater — presenting a balanced picture that obscures the directional reality: AI's "limitations" in generating alpha are its actual function — compressing margins and eliminating human labor — while its "potential" was always a transitional rent, now visibly decaying.

THE VERDICT

This paper is an autopsy with the patient still breathing.

It documents, with admirable data collection, the final innings of a temporary arbitrage window. The authors observe alpha compression, strategy differentiation, and adoption concentration, and conclude that AI has "both potential and limitations." What they are actually observing is the formation of the financial Sovereign tier — hedge funds that control AI capital — and the quiet pre-extinction of everyone else in the industry.

The paper is empirically useful as a lag indicator. The alpha decay it documents is evidence of AI deployment accelerating in financial services — one of the last cognitive white-collar domains to face full automation. When hedge fund alpha goes to zero, what remains is the Sovereign layer (who own the AI) and the rubble of the human analyst class.

No amount of academic rigor applied to a collapsing system makes the system less collapsed.