Sentient Compassion Values Corpus: Writings for AI Moral Reasoning

Moral-Circle-Alignment-Lab curated a dataset of writing modeling compassionate moral reasoning about nonhuman sentient beings. The corpus is designed for pretraining and fine-tuning language models to reason more carefully about decisions affecting sentient life. The dataset was last updated on May 28, 2026.

Use Cases

• Pretraining language models based on the corpus's curated writings on compassionate reasoning.
• Fine-tuning models for ethical decision-making based on examples concerning nonhuman sentient beings.
• Evaluating model alignment with compassionate values based on the described moral reasoning content.
• Training models to handle moral status uncertainty for entities like digital minds or insects.

Strengths

• The dataset is specifically curated for modeling compassionate moral reasoning, indicating a focused design.
• It covers a diverse range of sentient entities, including animals, insects, and digital minds.
• The dataset was last updated on May 28, 2026, suggesting recent maintenance.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Moral-Circle-Alignment-Lab

Collection Method

Curated collection of writings.

Freshness

Last updated 2026-05-28 07:13:14; freshness should be verified.