DeCoDe: Defer-and-Complement Decision-Making via Decoupled Concept Bottleneck Models

2025-05-25

AI Summary

This paper addresses dynamic task allocation in human-AI collaboration, proposing an interpretable ternary decision framework: autonomous AI execution, delegation to human experts, or human-AI complementary collaboration. Differing from conventional binary assignment methods, we introduce the first decoupled concept bottleneck model, incorporating a concept-level gating network and a surrogate loss function to enable fine-grained, adaptive, and transparent policy selection. Driven by interpretable concept representations, our architecture achieves high accuracy while significantly reducing human cognitive load and maintaining robustness under label noise and distributional shift—without compromising decision interpretability. Experiments on real-world datasets demonstrate consistent superiority over AI-only, human-only, and delayed-decision baselines across multiple metrics. The proposed framework establishes a novel paradigm for trustworthy human-AI collaboration in high-stakes domains.

Technology Category

• Humans and AI: Planning and Decision Support for Human-Machine Teams
• Philosophy and Ethics of AI: Accountability, Interpretability & Explainability
• Reasoning under Uncertainty: Sequential Decision Making

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

In human-AI collaboration, a central challenge is deciding whether the AI should handle a task, be deferred to a human expert, or be addressed through collaborative effort. Existing Learning to Defer approaches typically make binary choices between AI and humans, neglecting their complementary strengths. They also lack interpretability, a critical property in high-stakes scenarios where users must understand and, if necessary, correct the model's reasoning. To overcome these limitations, we propose Defer-and-Complement Decision-Making via Decoupled Concept Bottleneck Models (DeCoDe), a concept-driven framework for human-AI collaboration. DeCoDe makes strategy decisions based on human-interpretable concept representations, enhancing transparency throughout the decision process. It supports three flexible modes: autonomous AI prediction, deferral to humans, and human-AI collaborative complementarity, selected via a gating network that takes concept-level inputs and is trained using a novel surrogate loss that balances accuracy and human effort. This approach enables instance-specific, interpretable, and adaptive human-AI collaboration. Experiments on real-world datasets demonstrate that DeCoDe significantly outperforms AI-only, human-only, and traditional deferral baselines, while maintaining strong robustness and interpretability even under noisy expert annotations.

Problem

• Deciding AI, human, or collaborative task handling
• Overcoming binary AI-human choice limitations
• Enhancing interpretability in high-stakes AI decisions

Innovation

• Uses decoupled concept bottleneck models
• Implements three flexible collaboration modes
• Trains gating network with novel surrogate loss