# Morality is Contextual: Learning Interpretable Moral Contexts from Human Data with Probabilistic Clustering and Large Language Models

arXiv: https://arxiv.org/abs/2512.21439 · HTML: https://mlsanity.com/paper/2512.21439/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 30 references that are in the map.
- First posted on arXiv 24 Dec 2025.

## Builds on (most cited first, 18 of 18)

- [Discovering Language Model Behaviors with Model-Written Evaluations](https://mlsanity.com/paper/2212.09251/) (2022)
- [Eureka: Human-Level Reward Design via Coding Large Language Models](https://mlsanity.com/paper/2310.12931/) (2023)
- [Artificial Intelligence, Values and Alignment](https://mlsanity.com/paper/2001.09768/) (2020)
- [AI Alignment: A Comprehensive Survey](https://mlsanity.com/paper/2310.19852/) (2023)
- [Social Chemistry 101: Learning to Reason about Social and Moral Norms](https://mlsanity.com/paper/2011.00620/) (2020)
- [Evaluating the Moral Beliefs Encoded in LLMs](https://mlsanity.com/paper/2307.14324/) (2023)
- [Guiding Pretraining in Reinforcement Learning with Large Language Models](https://mlsanity.com/paper/2302.06692/) (2023)
- [Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs](https://mlsanity.com/paper/2502.17424/) (2025)
- [Qwen2.5-1M Technical Report](https://mlsanity.com/paper/2501.15383/) (2025)
- [Social Choice Should Guide AI Alignment in Dealing with Diverse Human Feedback](https://mlsanity.com/paper/2404.10271/) (2024)
- [The Moral Integrity Corpus: A Benchmark for Ethical Dialogue Systems](https://mlsanity.com/paper/2204.03021/) (2022)
- [Strong and weak alignment of large language models with human values](https://mlsanity.com/paper/2408.04655/) (2024)
- [The Moral Turing Test: Evaluating Human-LLM Alignment in Moral Decision-Making](https://mlsanity.com/paper/2410.07304/) (2024)
- [Zero-shot Model-based Reinforcement Learning using Large Language Models](https://mlsanity.com/paper/2410.11711/) (2024)
- [When Is It Acceptable to Break the Rules? Knowledge Representation of Moral Judgement Based on Empirical Data](https://mlsanity.com/paper/2201.07763/) (2022)
- [Exploring Qualitative Research Using LLMs](https://mlsanity.com/paper/2306.13298/) (2023)
- [Human Misperception of Generative-AI Alignment: A Laboratory Experiment](https://mlsanity.com/paper/2502.14708/) (2025)
- [Debunking Robot Rights Metaphysically, Ethically, and Legally](https://mlsanity.com/paper/2404.10072/) (2024)

## Built on by (most cited first, 1 of 1)

- [StereoTales: A Multilingual Framework for Open-Ended Stereotype Discovery in LLMs](https://mlsanity.com/paper/2605.10442/) (2026)

## Abstract (arXiv)

A key question in current AI alignment research is how to make AI algorithms learn moral values. Because human morality is highly context-dependent, actions are judged not only by their outcomes but by the context in which they occur. We present COMETH (Contextual Organization of Moral Evaluation from Textual Human inputs), a framework that integrates a probabilistic context learner with LLM-based semantic abstraction and human moral evaluations to model how context shapes the acceptability of ambiguous actions. We curate an empirically grounded dataset of 300 scenarios across six core actions relative to three moral rules (violating "Do not kill", "Do not deceive", and "Do not break the law") and collect ternary judgments (Blame/Neutral/Support) from N=101 participants. A preprocessing pipeline standardizes actions via an LLM filter and MiniLM embeddings with K-means, producing robust, reproducible core-action clusters. COMETH then learns action-specific moral contexts by clustering scenarios online from human judgment distributions using principled divergence criteria. To generalize and explain predictions, a Generalization module extracts concise, non-evaluative binary contextual features and learns feature weights in a transparent likelihood-based model. Empirically, COMETH roughly doubles alignment score with most human judgments relative to end-to-end LLM prompting (60% vs. 30% on average), while revealing which contextual features drive its predictions. The contributions are: (i) an empirically grounded moral-context dataset, (ii) a reproducible pipeline combining human judgments with model-based context learning and LLM semantics, and (iii) a more interpretable alternative than end-to-end LLMs for context-sensitive moral prediction and explanation.