All AI Models are Wrong, but Some are Optimal

2025-01-10

AI Summary

Conventional predictive AI models optimize for data fitting, yielding the most probable outcomes rather than directly supporting optimal sequential decisions—leading to suboptimal policies. Crucially, prediction accuracy does not guarantee decision optimality. Method: We establish a decision-centric modeling paradigm by formalizing necessary and sufficient conditions under which a predictive model supports an optimal policy. Leveraging rigorous integration of decision theory and statistical learning, we derive explicit mathematical mappings linking prediction error, model architecture, and policy optimality, and propose verifiable structural constraints on model design. Contribution: This work provides the first formal foundation for “decision-aware predictive modeling,” offering both theoretical guarantees and actionable design principles. It bridges the fundamental gap between predictive fidelity and decision efficacy, enabling predictive models to be explicitly optimized for downstream sequential decision-making tasks—not merely for pointwise accuracy.

Technology Category

• Reasoning under Uncertainty: Sequential Decision Making
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Other Foundations of Machine Learning

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

AI models that predict the future behavior of a system (a.k.a. predictive AI models) are central to intelligent decision-making. However, decision-making using predictive AI models often results in suboptimal performance. This is primarily because AI models are typically constructed to best fit the data, and hence to predict the most likely future rather than to enable high-performance decision-making. The hope that such prediction enables high-performance decisions is neither guaranteed in theory nor established in practice. In fact, there is increasing empirical evidence that predictive models must be tailored to decision-making objectives for performance. In this paper, we establish formal (necessary and sufficient) conditions that a predictive model (AI-based or not) must satisfy for a decision-making policy established using that model to be optimal. We then discuss their implications for building predictive AI models for sequential decision-making.

Problem

• AI prediction models
• decision-making optimization
• future forecasting limitations

Innovation

• Decision-making optimization
• Universal rules
• Continuous decision support