A systematic equity fund has deployed an AI system that generates position sizing recommendations for a $2B equity book. The head of portfolio construction is designing the human-AI decision boundary framework. The system performs well in normal market regimes but has limited back-test data covering liquidity crises. Which decision boundary design best preserves AI efficiency while maintaining appropriate human oversight for tail-risk scenarios?
Select an answer to reveal the explanation.
Short Explanation and Infographic
The best human-AI partnerships look like aviation autopilot: the autopilot handles cruise efficiently while the pilots retain full authority during takeoff, landing, and turbulence — the high-stakes moments where trained human judgment is irreplaceable. Regime-conditional autonomy applies exactly this logic: let the AI be fast and consistent in clear conditions, but build in the institutional equivalent of 'turbulence detected, human takes the wheel.' The kill-switch is always there; regime thresholds define when it gets handed over.
Full explanation below image
Full Explanation
The design of human-AI decision boundaries in investment management is a core challenge addressed by SR 11-7 (model risk management), the EU AI Act's risk-tiered framework, and emerging SEC guidance on automated trading systems. The core principle across all three regulatory frameworks is that autonomy should be commensurate with model confidence and environmental conditions, and that human accountability cannot be fully delegated to an automated system.
Regime-conditional autonomy (Option C) is the correct design because it resolves the fundamental tension between efficiency (which argues for more autonomy) and tail-risk management (which argues for more human oversight). The key insight is that the AI system's limitation is explicitly known: insufficient back-test data on liquidity crises. This is a documented model boundary, not a general deficiency. The appropriate response is to define a trigger — market stress indicators such as VIX spikes, bid-ask spread expansion, or funding liquidity metrics — that shifts the human-AI interface from 'approve exceptions' to 'approve executions.' The kill-switch at all times preserves the chain of human accountability that SR 11-7 requires.
Option A (autonomous with compliance triggers) creates the risk that compliance-level alerts fire after positions have already been taken in a stress scenario, which is precisely when reversing them is most costly. This is model risk management failure. Option B (human approval for all trades) eliminates the efficiency gains that justify AI deployment in position sizing at all — a $2B book with hundreds of positions cannot be human-approved for each rebalance. Option D (junior analyst buffer) introduces human review bottlenecks without adding the relevant expertise: junior analysts reviewing systematic signals in real time cannot add meaningful judgment on regime detection. The NIST AI RMF 'Manage' function specifically recommends dynamic autonomy levels calibrated to risk context, which aligns directly with Option C.