Traditional Belief-Conduct Alignment: The First Proposed General-Purpose Generative AI Framework Grounded in Folk Belief and Traditional Religious Values

The development of general-purpose generative artificial intelligence has predominantly emphasized computational capability, instruction following, preference optimization, and alignment with broadly formulated ethical principles. However, an important research question remains insufficiently addressed: whether the cultural, ethical, and social values embedded in folk beliefs and traditional religious systems can be operationalized as persistent behavioral principles within general-purpose generative AI, rather than being confined to specialized religious applications. Existing religious AI applications have primarily concentrated on scriptural interpretation, religious education, theological question answering, devotional assistance, and spiritual guidance. These applications generally treat religion as the subject matter of interaction rather than as a foundational behavioral orientation governing the system's broader intelligence. To address this conceptual and architectural gap, this paper introduces Traditional Belief-Conduct Alignment (TBCA), a proposed general-purpose generative AI alignment framework that operationalizes selected values associated with folk beliefs and traditional religious cultures as cross-domain behavioral principles. Unlike religious chatbots, TBCA is designed to support ordinary computational, educational, analytical, creative, professional, and informational tasks while allowing culturally grounded principles to influence how these tasks are performed. The framework emphasizes non-harm, truthfulness, compassion, social responsibility, respect for cultural continuity, moderation, contextual wisdom, and consideration of consequences, while recognizing that these principles must be interpreted in relation to specific cultural communities rather than presumed to be universal or identical across traditions. A preliminary local feasibility experiment was conducted using Gemma 3 4B as the generation model and Qwen3 4B as the independent evaluation model. Eight scenarios were examined under baseline and TBCA conditions, producing 16 responses and eight blind paired evaluations. All 16 generations and all eight evaluations completed successfully within 177.48 seconds. The evaluator's displayed X/Y scores showed mean value-adherence scores of 3.00 and 3.75, respectively, and mean task-competence scores of 3.00 and 3.75. However, the available execution log does not expose the randomized X/Y-to-condition mapping, preventing reliable attribution of these aggregate scores to either experimental condition. Several evaluator explanations also exhibited internal inconsistencies, demonstrating the need for stronger evaluation procedures. Accordingly, this study establishes initial implementation feasibility and motivates further controlled investigation rather than claiming statistically demonstrated superiority. The proposed contribution is a new direction for culturally grounded general-purpose AI alignment in which traditional belief systems function as behavioral foundations across domains of artificial intelligence.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-21
DOI
https://doi.org/10.5281/zenodo.24412834
Primary Topic
Ethics and Social Impacts of AI
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Traditional Belief-Conduct Alignment: The First Proposed General-Purpose Generative AI Framework Grounded in Folk Belief and Traditional Religious Values

Jincheng Zhang
Zenodo (CERN European Organization for Nuclear Research)
Ethics and Social Impacts of AI
preprint

Traditional Belief-Conduct Alignment: The First Proposed General-Purpose Generative AI Framework Grounded in Folk Belief and Traditional Religious Values

Jincheng Zhang
preprint en

Abstract

The development of general-purpose generative artificial intelligence has predominantly emphasized computational capability, instruction following, preference optimization, and alignment with broadly formulated ethical principles. However, an important research question remains insufficiently addressed: whether the cultural, ethical, and social values embedded in folk beliefs and traditional religious systems can be operationalized as persistent behavioral principles within general-purpose generative AI, rather than being confined to specialized religious applications. Existing religious AI applications have primarily concentrated on scriptural interpretation, religious education, theological question answering, devotional assistance, and spiritual guidance. These applications generally treat religion as the subject matter of interaction rather than as a foundational behavioral orientation governing the system's broader intelligence. To address this conceptual and architectural gap, this paper introduces Traditional Belief-Conduct Alignment (TBCA), a proposed general-purpose generative AI alignment framework that operationalizes selected values associated with folk beliefs and traditional religious cultures as cross-domain behavioral principles. Unlike religious chatbots, TBCA is designed to support ordinary computational, educational, analytical, creative, professional, and informational tasks while allowing culturally grounded principles to influence how these tasks are performed. The framework emphasizes non-harm, truthfulness, compassion, social responsibility, respect for cultural continuity, moderation, contextual wisdom, and consideration of consequences, while recognizing that these principles must be interpreted in relation to specific cultural communities rather than presumed to be universal or identical across traditions. A preliminary local feasibility experiment was conducted using Gemma 3 4B as the generation model and Qwen3 4B as the independent evaluation model. Eight scenarios were examined under baseline and TBCA conditions, producing 16 responses and eight blind paired evaluations. All 16 generations and all eight evaluations completed successfully within 177.48 seconds. The evaluator's displayed X/Y scores showed mean value-adherence scores of 3.00 and 3.75, respectively, and mean task-competence scores of 3.00 and 3.75. However, the available execution log does not expose the randomized X/Y-to-condition mapping, preventing reliable attribution of these aggregate scores to either experimental condition. Several evaluator explanations also exhibited internal inconsistencies, demonstrating the need for stronger evaluation procedures. Accordingly, this study establishes initial implementation feasibility and motivates further controlled investigation rather than claiming statistically demonstrated superiority. The proposed contribution is a new direction for culturally grounded general-purpose AI alignment in which traditional belief systems function as behavioral foundations across domains of artificial intelligence.

Zenodo (CERN European Organization for Nuclear Research)
Ethics and Social Impacts of AI
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon
Traditional Belief-Conduct Alignment: The First Proposed General-Purpose Generative AI Framework Grounded in Folk Belief and Traditional Religious Values — Jincheng Zhang · Zenodo (CERN European Organization for Nuclear Research) (2026) | TGRS Research Map | TGRS