A traditional hedge fund is run by human portfolio managers who use research, models and judgment to decide what to trade. An "AI hedge fund" hands the final decision to a model — anything from a statistical algorithm to a large language model (LLM) like Claude, GPT or Grok — that weighs the same inputs and outputs a decision.
The important distinction is where the decision is made. Many funds already use AI and machine learning to generate signals; the model is a tool feeding a human. In an AI-run book, the model itself makes the call, and the human role is limited to building the system and enforcing rules.
Fully autonomous AI funds managing large pools of public money are still rare. Most visible examples are paper (simulated) portfolios or internal research experiments, because performance is unproven and regulation around AI-driven advice is strict.
Whether an AI can consistently beat the market is an open question. Markets are highly competitive, and an edge that is easy to describe tends to get arbitraged away. The honest answer is that it has not been convincingly demonstrated over long, live track records.