analiza aiqbench
This research introduces COMETH, a framework that significantly improves AI alignment by modeling how context influences moral judgments rather than relying on opaque end-to-end LLM prompting. By integrating probabilistic clustering with human evaluations, the system achieves roughly double the alignment score of standard prompting while offering interpretable insights into which contextual features drive specific moral predictions. This approach addresses a critical gap in making AI systems more transparent and adaptable to nuanced ethical scenarios.
Morality is Contextual: Learning Interpretable Moral Contexts from Human Data with Probabilistic Clustering and Large Language Models
arXiv:2512.21439v2 Announce Type: replace-cross Abstract: 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.
Abstract: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.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2512.21439 [cs.CL]
(or arXiv:2512.21439v2 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2512.21439 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 341 cuvinte
Subiecte conexe
The basic argument for making AI better at moral philosophy
LessWrong · 14 oct. 2026
How might continual learning affect safety and alignment?
LessWrong · 13 iun. 2026
The J-Space Debate, Agent Swarms, and Pacing Frontier AI - Digital Minds Newsletter #4
LessWrong · 18 sept. 2026
The user in the LLM might be a moral patient too (a strange implication of the PSM and simulators)
LessWrong · 15 oct. 2026