Research map / Large language models / AI and society
Social, Legal, Ethical, Empathetic, and Cultural Rules: Compilation and Reasoning
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.
Translates SLEEC (social, legal, ethical, empathetic, cultural) normative rules into classical propositional logic via a formal 'unless ... in which case' semantics, analyses reasoning complexity, and shows implementation in ASP and Prolog.
Problem
SLEEC rules elicited in natural language for AI and autonomous systems need systematic translation into a machine-interpretable formal language supporting automated reasoning.
Approach
A linguistic analysis of the SLEEC rule pattern justifies defining 'unless' as a material implication (¬ψ→ϕ) and 'unless ... in which case' as a conjunction, yielding a quadratic-size classical logic compilation; reasoning is then done with SAT solvers, ASP, or Prolog.
Key results
- Proves that entailment of an obligation from SLEEC rules is coNP-complete even when conditions and obligations are literals, via reduction from 3SAT.
- Shows entailment is in PTIME when all conditions are negative literals and all obligations positive literals, by reduction to HORNSAT.
- Shows the compiled form of a SLEEC rule has size quadratic in the original rule size (|compile(σ)| = O(|σ|^2)).
Surprising findings
- 'Unless' is not equivalent to exclusive-or; the converse implication is only a Gricean conversational implicature, not part of the meaning.
Limitations
- Worst-case reasoning is coNP-hard unless restricted to Horn-like forms.
- Ethical review of the overall system remains essential beyond the formal implementation.
- Demonstrated on a single running example (nursing home curtain rule).
Experimental setup
Baselines: SLEECVAL normative multi-agent formalisms (Meyer and Wieringa 1994; Gabbay et al. 2013, 2
Builds on: SLEEC rule elicitation methodology (Townsend et al. 2022) Quine's logical reading of unless SLEECVAL (Yaman et al. 2023)
Closest papers by meaning
- RULEBREAKERS: Challenging LLMs at the Crossroads between Formal Logic and Human-like Reasoning (ICML 2025 · similarity 0.77)
- AI Toolkit: Libraries and Essays for Exploring the Technology and Ethics Behind AI (AAAI 2025 · similarity 0.77)
- Regulating AI: Applying Insights from Behavioural Economics and Psychology to the Application of Article 5 of the EU AI Act (AAAI 2024 · similarity 0.77)
- OlympicArena: Benchmarking Multi-discipline Cognitive Reasoning for Superintelligent AI (NeurIPS 2024 · similarity 0.77)
- Beyond Right and Wrong: Evaluating Second-order Social Reasoning in Large Language Models (NeurIPS 2026 · similarity 0.77)
- Counterfactual Metarules for Local and Global Recourse (ICML 2024 · similarity 0.77)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
