Research map / Large language models / AI and society

Towards Computational Foreseeability

AAAI 2025 · Large language models › AI and society · approach: Human-centered trust and agency

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Position paper advocating a computational foreseeability framework for autonomous vehicles based on the legal Hand/BPL formula (burden < probability × loss), with worked examples and open research challenges.

Problem

Autonomous systems, especially AVs, lack an accountable, legally grounded way to decide what counts as foreseeable and acceptable behavior in safety-critical scenarios; current data-driven foreseeability bounds are rigid, ignore uncertainty, and rely on simulation data.

Approach

Adapt the legal BPL (Hand) formula to computationally establish foreseeability: compute burden as the reward gap between a real AV and a reasonable-AV standard, probability from reachable-state failure likelihood, and loss from cost tables (e.g., NSC crash costs), then compare burden to probability times loss.

Key results

• Worked neighborhood example yields burden 28,000 versus combined probability-loss 1,162,965, judged foreseeable.
• Highway example yields burden 1,878,000 versus combined probability-loss 868,190, judged unforeseeable.
• The paper presents no empirical evaluation; the numbers are illustrative calculations.

Limitations

• No implemented system or empirical validation; illustrative only.
• Unresolved challenges in defining reasonable AV standards, burden, and probability estimation.
• Loss values depend on contested monetary valuation of human life and lack ground truth for evaluation.

Experimental setup

Datasets: National Safety Council economic cost of motor vehicle crashes (2022)

Baselines: data-driven foreseeability bounds cost minimization reward maximization

Builds on: Hand formula (BPL) scenario-based verification opportunity cost

Closest papers by meaning

• From Gambits to Assurances: Game-Theoretic Integration of Safety and Learning for Interactive Robotics (AAAI 2025 · similarity 0.86)
• Military AI Needs Technically-Informed Regulation to Safeguard AI Research and its Applications (NeurIPS 2025 · similarity 0.83)
• A Physics-Informed Machine Learning Framework for Safe and Optimal Control of Autonomous Systems (ICML 2025 · similarity 0.83)
• Position: AI Safety Must Embrace an Antifragile Perspective (ICML 2025 · similarity 0.83)
• Hybrid-Driving: An Autonomous Driving Decision Framework Integrating Large Language Models, Knowledge Graphs and Driving Rules (AAAI 2025 · similarity 0.83)
• STL: Still Tricky Logic (for System Validation, Even When Showing Your Work) (NeurIPS 2024 · similarity 0.82)